J. Even Ø. Nilsen, Helge Drange, Kristin Richter, Eystein Jansen and Atle Nesje :

Changes in past, present, and future sea level on the coast of a project by Nansen Environmental and Remote Sensing Center and UNI Research, at the Bjerknes Centre for Climate Research, funded by the City of , Department of Urban Development, Climate, and Environmental A airs.

NERSC Special Report no. 89 Bjerknes Centre for Climate Research publication no. R101

Bergen, August 2012 This is NERSC Special Report 89, publication no. R101 from the Bjerknes Centre for Climate Research.

The main parts of this report are to be cited as follows.

Chapter 4 is to be cited as: Jansen, E. (2012). Paleoclimatic perspectives on sea level. In Nilsen, J.E.Ø., H. Drange, K. Richter, E. Jansen, A. Nesje. Changes in the past, present, and future sea level on the coast of Norway. NERSC Special Report 89, Bergen, Norway.

Chapter 5 is to be cited as: Richter, K., J.E.Ø. Nilsen, H. Drange (2012). Contributions to observed sea level change for1960-2010. In Nilsen, J.E.Ø. , H. Drange, K. Richter, E. Jansen, A. Nesje. Changes in the past, present, and future sea level on the coast of Norway. NERSC Special Report 89, Bergen, Norway.

Chapter 6 is to be cited as: Drange, H., J.E.Ø. Nilsen, K. Richter, A. Nesje (2012). Updated estimates of future sea level rise on the Norwegian coast. In Nilsen, J.E.Ø., H. Drange, K. Richter, E. Jansen, A. Nesje. Changes in the past, present, and future sea level on the coast of Norway. NERSC Special Report 89, Bergen, Norway.

The published article in Appendix 2 is to be cited as: Richter, K., J.E.Ø. Nilsen, H. Drange (2012). Contributions to sea level variability along the Norwegian coast for 1960-2010. J. Geophys. Res., 117, doi:10.1029/2009JC007826.

The full report is to be cited as: Nilsen, J.E.Ø., H. Drange, K. Richter, E. Jansen, A. Nesje. (2012). Changes in the past, present, and future sea level on the coast of Norway. NERSC Special Report 89, Bergen, Norway. 48 pp.

City of Bergen, Department of Urban Development, Climate, and Environmental A airs: www.bergen.kommune.no/byutvikling Nansen Environmental and Remote Sensing Center: www.nersc.no Uni Research: www.uni.no Bjerknes Centre for Climate Research: www.bjerknes.uib.no University of Bergen: www.uib.no MARE: www.mare-project.eu August 2012

Changes in the past, present, and future sea level on the coast of Norway1

Project report to the City of Bergen, Department of Urban Development, Climate, and Environmental Affairs

1,2 3,2 4,2 5,4,2 5,2 Jan Even Øie Nilsen , Helge Drange , Kristin Richter , Eystein Jansen and Atle Nesje 1 Nansen Environmental and Remote Sensing Center, Bergen 2 Bjerknes Centre for Climate Research, Bergen 3 Geophysical Institute, University of Bergen 4 Uni Research AS, Bergen 5 Department of Earth Science, University of Bergen

Sea levels are rising, predominantly due to the warming of the oceans, melting of land- based ice, and ground water depletion. In addition land surfaces rise and sink. The west coast of Norway is still rising after the retreat of the Fennoscandian ice sheet of the last ice. Presently, the rates of ocean and land rise are comparable, but under global warming the sea levels on the Norwegian coast are expected to rise by 20 to 80 cm by the end of the century. In 50 years about half of this rise is estimated. In the latter half of this century the expected sea level rise will impose increased challenges upon existing infrastructure, and adapting plans for new infrastructure to an ever-rising sea level can be advantageous. In this research project, changes in sea level in prehistoric times as well as during the latest 50 years are studied, the state of the present sea level is assessed, and updated estimates for sea level rise in the 21st century is presented.

1 To be cited as Nilsen, J.E.Ø., H. Drange, K. Richter, E. Jansen, A. Nesje (2012). Changes in the past, present, and future sea level on the coast of Norway. NERSC Special Report 89, Bergen, Norway. 48 pp.

1 August 2012

Table of Contents 1. Summary...... 3 2. Administrative details...... 4 3. Introduction...... 4 4. Paleoclimatic perspectives on sea level ...... 5 Mean sea level ...... 5 Rates of change ...... 6 5. Contributions to observed sea level change for 1960-2010...... 8 Methods ...... 9 Results...... 10 Discussion...... 12 Conclusions ...... 13 6. Updated estimates of future sea level rise on the Norwegian coast ...... 14 Background ...... 14 Observed sea level rise...... 15 Global sea level in the future...... 15 Regional sea level in the future ...... 17 7. Dissemination ...... 21 Bibliography ...... 22 Appendix...... 24 Appendix 1: Estimates of future sea level rise for the Norwegian coastal municipalities...... 24 Appendix 2: Peer review publication on sea level change during the past 50 years .31

2 August 2012

1. Summary The estimated global sea level rise for the two recent decades is 3 mm/yr, twice as fast as the average rise throughout the last century. We know the surface of the oceans will continue to rise for a long time into the future, even hundreds of years after humanity learns how to control greenhouse gas emissions. In this project, prehistoric sea level and rates of change have been assessed, the mechanisms involved in sea level rise have been studied based on observations from modern times, and future sea level rise has been estimated based on the current knowledge. About 3 million years ago, when the continents were already in today's positions, the climate on Earth was significantly warmer than today, the large ice sheets of Greenland and Antarctica were smaller and the sea level 10-30 m higher. During the last interglacial (about 120.000 years ago) global temperature was about 1-2°C warmer than today. The sea level was 4-10 m higher, mainly due to less water stored as ice on land and the thermal expansion of the oceans. The sea level change during the last interglacial was around 2 mm/yr, which is comparable to the presently observed rates. If today's ice sheets of Greenland and West-Antarctica were to become unstable and partly collapsing, sea level rates may become similar to those found after the ice ages. These rates have been estimated to be up to 40 mm/yr at certain locations. In modern times, the different processes affecting regional sea level change can be studied using various observations. In the study focussing on the Norwegian coast, it is shown that short-term changes in local sea level are to a large extent caused by changes in temperature (thermal expansion) and salt content (haline contraction) of seawater, as well as changes in local atmospheric pressure. In contrast, less than half of the observed long-term changes (the trend) can be explained by these processes and land uplift. In fact, apart from the land uplift, only thermal expansion contributes to a significant trend along the entire Norwegian coast. The observed trend in relative sea level (the sea level observed from shore) is 0.9 mm/yr in Bergen in the period 1960-2010. For the absolute sea level (i.e. without compensating for land uplift) the rise would be 2.6 mm/yr. Of this 0.9 mm/yr can be attributed to thermal expansion, and 0.7 mm/yr is estimated to be due to melting land ice. The remainder is subject to different processes with large uncertainties, and further research is necessary to accurately quantify their importance. Future sea level rise can be estimated by combining sea level due to changes in the oceans' temperature, salt content and circulation as projected by climate models, with estimated contributions from land ice and water stored on land, changes in the gravity field, redistribution of sea water within the oceans and vertical land movement. Based on estimates of future global sea level rise and by taking into account the aforementioned processes and standard estimates for uncertainties we estimate that the sea level in Bergen will be between 20 to 80 cm higher within 100 years, with a probability of 66%. The data needed to compute these numbers are taken from the latest available literature. The computation of the likely sea level change has been conducted for all coastal communities in Norway (Figure 5 and Table 4). Due to the increasing amount and higher accuracy of observations and a constantly increasing understanding of the processes affecting sea level, projections of global and local sea level rise should be updated regularly.

3 August 2012

2. Administrative details Project: Project number M 760, Sub-project M780: Changes in the past, present, and future sea level on the coast of Norway. The project is a sub-project under MARE. Administrative responsible: Prof. Ola. M. Johannessen, Nansen Environmental and Remote Sensing Center (NERSC) Prof. Eystein Jansen, Bjerknes Centre for Climate Research (Uni Research AS) Research leadership: Dr. J. Even Ø. Nilsen (NERSC) Prof. Helge Drange (University of Bergen) Contract partner: Nansen Environmental and Remote Sensing Center Thormøhlensgt. 47 N-5006 Bergen Norway Duration: May 2009 - May 2012

3. Introduction Tide gauges around the world show that global sea level has risen by about 17 cm during the past 100 years, and it appears that the sea level rise is accelerating. Satellite measurements indicate a rise of 3 mm/yr since 1993 or almost twice as much as the average rise during the past century. It is also well known that the rise will continue even after CO2 emissions will be stabilized. It is therefore necessary to adapt and be prepared to rising sea levels. The degree to which measures should be taken in the present and future to mitigate the consequences of rising seas depends on the risk assessment. This is not addressed in this report. The projection of sea level rise in the future is subject to large uncertainties. These uncertainties arise both from uncertainties in future greenhouse gas emissions as well as uncertainties in the relative contributions from the different processes leading to changing sea levels. E.g. how quickly the ocean is warming, how fast the ice on land melts, what is the rate of vertical land movement, and how is Earth's gravity field affected? The structure of the project has been as follows: Prehistoric data has been studied to assess previous sea levels, as well as constraining possible rates of change; the instrumental record, the present, has been used to study sea level changes along the Norwegian coast and its causes during the past 50 years; projections of future sea level have been computed using a combination of published data and increased understanding of the individual contributions.

4 August 2012

4. Paleoclimatic perspectives on sea level 2 E. Jansen Due to the commonly existing sea level changes in Earth´s history, present and future changes might be placed into a longer time perspective from knowledge about past sea level changes. Of interest here are both the absolute values of sea level changes in past warm periods of Earth´s history and the rates of change experienced during previous periods of rapid sea level change. Sea level in previous periods can be calculated from data that reconstructs the position of earlier shorelines, sedimentary basins which change character from marine to lake conditions, or through the utilisation of fossils from organisms that have a known habitat at the sea surface, e.g. corals, and therefore can trace the mean sea level of the time they lived. When it is possible to date the mean sea level position in situations such as these in a specific locality, one needs to further correct sea level estimates to modern sea levels due to vertical movements of the land surface due to isostatic movements or tectonic factors that have changed the position of the locality. Such corrections can in many places be made with high degree of certainty, hence it is possible to establish both the absolute sea level of past times and, in some situations, also the rates of change of past sea level variations. Mean sea level Due to the importance of sea level in the context of global warming, a high number of new studies of past sea levels have been performed during recent years, and a growing body of scientific literature sheds light on these changes. Sea level in the Pliocene epoch During the Pliocene era (3-5 million years before present) the mean climate state of the Earth was over long periods (more than 100.000 years) significantly warmer than present. At the same time the CO2-content of the atmosphere is estimated to have been 400-420 ppmv (parts per million by volume), i.e. not significantly higher than the concentration will be in one or two decades from now. This period is not so far back in time and the surface of the Earth was quite similar to present in terms of the positioning of continents and size and place of mountain ranges. Since the warm periods of the Pliocene had such a long duration, it is reasonable to assume that the slow acting elements of the climate system were in equilibrium during this period, which therefore can provide us with estimates of future sea level on the longer-term with a slow and steady adjustment to the existing boundary conditions. Recent literature indicates that Pliocene sea levels were 20 ± 10 m above modern sea levels, with a 66% likelihood that it was between 12 and 30 higher than today, based on a set of 34 globally distributed locations. These estimates are slightly lower than earlier estimates (which were up to 40 m higher than today; Brigham-Grette and Carter, 1992). The adjustment to lower estimates is due to more precise modelling of the movements of the Earth's crust at the locations. These results imply that the land-ice volume in Antarctica and Greenland were significantly less

2 To be cited as Jansen, E. (2012). Paleoclimatic perspectives on sea level. In Nilsen, J.E.Ø., H. Drange, K. Richter, E. Jansen, A. Nesje. Changes in the past, present, and future sea level on the coast of Norway. NERSC Special Report 89, Bergen, Norway.

5 August 2012 in the Pliocene than now, and that the West-Antarctic ice-sheet probably was not existing during periods of this time (Naish et al., 2009; Pollard and DeConto, 2009). The last interglacial The last interglacial period, approximately 130.000-120.000 years before present, was a period with mean global temperature 1-2oC above late 20th century temperatures, and a significantly enhanced warming in polar regions. At this time the Earth's orbit brought the Earth closer to the sun during Northern Hemisphere summer than now with a slightly warmer climate as a consequence. Atmospheric CO2 levels were, however, at 280-300 ppmv, i.e. significantly lower than they are now. There are extensive data sets available to document that mean sea levels were substantially higher than now during the last interglacial. The newest estimates indicate that sea level was at least 6 m higher than now (with an uncertainty interval spanning 4 to 10 m above modern sea level), based on lifted marine terraces and the oxygen isotope composition of the sea (which change when global land-ice volume changes) (Kopp et al., 2009; Rohling et al., 2009; Lisiecki and Raymo, 2005). Both Greenland and West Antarctic ice sheets were smaller than now, likely 30% smaller in the case of Greenland. Estimates of the steric effect of the warmer ocean indicate at most 0.3 m higher sea level from this effect (McCay et al., 2011). Most of the sea level rise must therefore be ascribed to deglaciation of the ice sheets. It is difficult to discriminate between Greenland and West Antarctica, and this issue is an object of intensive research. These results show, however, that polar ice sheets and associated sea levels are highly sensitive to increasing temperatures. Present interglacial period (the Holocene) After sea levels rose about 120 m at the end of the last glaciation, ending at about 6000 years before present, due to the demise of the large continental ice sheets, global sea levels have been quite stable. During the last 5000 years sea levels have been relatively constant. Some long-term adjustments of Antarctic ice sheets have been registered to last as long as up to 3000 years before now, hence one can only compare present sea levels to historic sea levels in detail over the last 2000 years. Our methods to reconstruct sea levels for this period have an accuracy of about 20 cm. Until approximately 1900 AD sea levels fluctuated within a range of 0 to 20 cm, while after 1900 AD these fluctuations were replaced by a monotonous rise to modern sea levels. Rates of change The accuracy of dating past sea levels is too low to allow estimates of rates of change for older times than the last interglacial. A number of studies estimate how fast sea level rose up to the highest recorded levels of the last interglacial. These indicate a maximum rate of change of 1.6 m/100 years (Rohling et al., 2009), but the results are disputed. Other estimates conclude with maximum rates 1/10 of this, i.e. 20 cm/100 years (Blanchon et al., 2009). The present day rate of change is, for comparison, 30 cm/100 years. During deglaciations at the end of the ice ages, large continental ice sheets collapsed, and the rates of change of sea level during these most extreme periods of sea level rise may indicate an upper boundary for how fast sea level may rise in a situation where the Greenland and West Antarctic ice-sheets may enter into a similar situation of collapse. Reconstructed rates of change of up to 4 m/100 years have been estimated for shorter periods of the last deglaciation rate (Bard et al., 1990; Hanebuth et al. 2000). It is difficult to provide global estimates during these periods due to many local imprints, but it is

6 August 2012 reasonable to believe that the rates may have been at least 2 m/100 years. Such a rate can be extracted from the reconstruction from Sotra by Lohne et al. (2007; Figure 1). After extracting the imprint from isostatic rise of the land surface due to the disappearance of the glacial ice load, the mean sea level rise for the initial phase of the deglaciation was approximately 60 cm/100 years.

Figure 1: Relative sea level curve for Sotra, based on studies of lake sediments from isostatically raised lakes, during the past 14.500 years. Crosses represent radiocarbon dates with uncertainty estimates in time and height, which the curve is drawn through. The dotted line is drawn through the actual dates, while the stippled line is considered more accurate due to probable dating errors imposed by the Storegga-tsunami 8.100 years ago. Figure is from Lohne et al. (2007).

7 August 2012

5. Contributions to observed sea level change for 1960-20103 K. Richter, J.E.Ø. Nilsen, H. Drange Global sea level has been rising by about 20 cm during the last century and is expected to continue to rise in the 21st century. The rise and variability is not spatially uniform. To be able to project local changes in relative sea level (RSL), it is important to identify the processes that govern regional RSL variability. In this study, we assess the importance of different contributions to RSL variability along the coast of Norway in the period 1960– 2010. This work is published as a research article in Journal of Geophysical Research (Richter et al., 2012; Appendix 2). The following is a summary of the work and the most important results. The relative importance of the different factors that contribute to changes in global, and local, sea level varies with time. According to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC), thermal expansion contributed with about 40% while glaciers and ice sheets accounted for the remaining 40% and 20%, respectively, in the period 1961–2003 (Meehl et al., 2007). For 2003-2007 thermal expansion and ice sheets have been estimated to contribute 10% and 40%, respectively (Cazenave and Llovel, 2010). The individual contributions also vary geographically, that is their magnitude may be different in different parts of the globe. The focus of this study is on the Norwegian coast and the following effects are investigated: the expansion and contraction of sea water due to changes in temperature and salt content, changes in air pressure and the effect of land uplift. Changes in the heat (temperature) and salt (salinity) content of sea water result in thermal expansion and haline contraction. The resulting change in sea level is referred to as change in steric height. Warmer water expands while more salt leads to contraction of seawater. In the world oceans the temperature effect dominates, but in the cold waters of the Nordic Seas and Arctic Ocean changes in salinity are at least as important as temperature effects. Steric height variations alter the sea level by changing the volume of seawater through expansion and contraction in contrast to adding water to or removing it from an ocean region. The latter can be due to melting land ice (glaciers and ice sheets) or moving of water masses from one ocean region to another. Changes in atmospheric pressure over the oceans will cause sea water to move and therefore change the local sea level. A low-pressure system means less air masses (less weight) and the ocean surface will respond by rising. The water necessary to do so will come from areas of high air pressure where more weight depresses the ocean surface. This way, air pressure changes move water from areas of high atmospheric pressure to low atmospheric pressure. As a rule of thumb, a 1 mbar decrease in air pressure results in a 1 cm rise in sea level. In addition, a low-pressure system (a storm) is usually accompanied by winds that, through friction, get the seawater moving. The Norwegian coast is commonly subject to incoming storms from the southwest that are pushing water towards the shore. The results are storm surges for isolated storms and long-term increased sea level for a continuous stream of storms from the southwest.

3 To be cited as Richter, K., J.E.Ø. Nilsen, H. Drange (2012). Contributions to observed sea level change for 1960-2010. In Nilsen, J.E.Ø., H. Drange, K. Richter, E. Jansen, A. Nesje. Changes in the past, present, and future sea level on the coast of Norway. NERSC Special Report 89, Bergen, Norway.

8 August 2012

Land uplift is an important factor when assessing sea level changes. It is the sea level with respect to the shore that is of major interest for adaptation and mitigation. In addition direct measurements of sea level are being carried out from land by tide gauges. Therefore, we separate between absolute sea level (sea surface height, SSH) and relative sea level (RSL), where the latter is relative to the shore. The land uplift in Norway is mainly due to the complete melting of the Fennoscandian ice sheet that covered Scandinavia during the last ice age. The Earth is still adjusting to the removal of the weight of the ice masses (glacial isostatic adjustment, GIA) and, in the period this study is concerned with, the effect can be described as a constant uplift. Methods In this study, we consider changes induced by atmospheric and thermohaline variability, as well as vertical land uplift. The combination of these contributions yields the reconstructed RSL,

RSLrc = ηp +ηT +ηS + GIA , (1)

which will be compared to the observed RSL. In the expression above, ηp is the SSH variability due to surface air pressure fluctuations, ηT and ηS are the thermosteric and € halosteric contributions, respectively, and GIA is a linear trend representing vertical land uplift due to glacial isostatic adjustment. Accordingly,

RSL = RSLrc +ηres, (2)

where ηres is the sea level residual that is not explained by our reconstruction.

72 o € N

Ingøy Honningsvåg

Hammerfest

68 o N Harstad Tromsø Eggum Kabelvåg Skrova Bodø

64 o N Rørvik

Bud Heimsjø Ålesund Måløy 60 o Sognesjøen N Bergen Oslo Ytre Utsira Indre Utsira Stavanger Lista Tregde o 56 N

o o o 0 10 E 20 E Figure 2: Positions of the tide gauges used in the analysis (black) and locations of the hydrographic stations (blue). The observed RSL is obtained from tide gauges along the entire Norwegian coast (Figure 2). Due to the limited length of the available records the study focuses on the period 1960-

9 August 2012

2010. We use monthly data in our analysis and perform the reconstruction according to equation 1 for each individual tide gauge. Atmospheric pressure is obtained from the reanalysis data of the National Centers for Environmental Prediction-National Center for Atmospheric Research (NCEP-NCAR). The data is from an atmospheric model that includes information from actual meteorological observations (including local air pressure) and provides a consistent estimate of atmospheric variables (pressure, wind, etc.) at positions in between the meteorological stations. The spatial resolution is 2.5 degrees and we use the pressure from the ocean grid point closest to each tide gauge. For components of steric height we use hydrographic station data along the Norwegian coast provided by the Institute of Marine Research (IMR), Bergen, Norway. There are eight permanent stations along the Norwegian coast (Figure 2) that have been maintained for several decades and provide vertical profiles of temperature and salinity throughout the year and through the whole water column. The locations of the hydrographic stations are not identical to the locations of tide gauges (Figure 2). Therefore, RSL observations from tide gauges have been paired with the steric height based on their location and the highest correlation coefficients between steric height and RSL observed with tide gauges (see Appendix 2 for details). Land uplift (GIA) can be estimated in different ways by including geodetic datums (reference points on land), global positioning system (GPS) measurements, geodynamic modelling, or sea level observations. The estimates of the group around Peltier (Peltier, 2004) are widely used. However, they apply a global Earth model, such that small-scale anomalies in Earths structure are not properly modelled and local uplift rates cannot be expected to be accurate. We therefore chose to use uplift rates by Vestøl (2006) who combined levelling (geodetic datums), historical tide gauge recordings, and GPS data to derive land uplift rates for Fennoscandia. In the meantime, a new and improved data set has been released by Simpson et al. (2012). We will discuss our results with respect to the new rates and discuss the differences in the next section. Results The goal of the study was to assess how well the combination of the chosen components represents the observed sea level trend and variability for the past 50 years. Covariance, that is the common variability on time scales shorter than the period that is studied, reveals the direct influence different processes have on observed sea level. This study shows that the three contributions (thermal expansion, haline contraction and atmospheric pressure changes) are responsible for 70-85% of the observed sea level variability at all positions except the two southernmost stations (Tregde and Oslo). Land uplift is not considered in the covariance analysis, as it is a constant trend. For Bergen, 76% is explained. These are high numbers when taking into account that all data are also subject to random noise. The high numbers show that a large part of the variability is explained by our relatively simple reconstruction (equation 1). For Bergen, 46% of the observed RSL variability is explained by the pressure effect, 29% by the thermosteric effect (sea temperature) and 34% by the halosteric effect (salt). As these contributions are not completely independent from each other, their combination explains less than the sum of the individual components. The explained variances are somewhat reduced when using yearly data indicating that other processes become more important on longer time scales.

10 August 2012

Honningsvåg Hammerfest Tromsø Harstad Narvik Kabelvåg Bodø Rørvik Heimsjø Kristiansund Ålesund RSL Måløy Bergen SSH Stavanger SSH new Tregde Oslo

−2 −1 0 1 2 3 4 −1 trend (mm yr ) Figure 3: Trends in the sea level at different locations on the Norwegian coast, with error bars. In black is the relative sea level (RSL) directly from tide gauge records, in red is our estimate of absolute sea level rise (i.e. corrected for land uplift), while in green is absolute sea level rise using the land uplift estimates from Simpson et al. (2012).

The trends in sea level during the period 1960-2010 vary geographically (Figure 3). This is in particular true for the RSL trends (the trends observed from the shore) and is due to the uplift rates being very different from one station to another. The uplift rate at a given point depends on the distance of that point to the centre of mass of the Fennoscandian ice sheet, the Gulf of Bothnia. The RSL trends are positive from Tregde to Ålesund and north of Harstad, and negative or close to zero elsewhere. After correcting for land uplift and computing the SSH trends, the differences are reduced and the trends are positive and larger than 1.7 mm/yr at all stations (the mean in Figure 3 is 2.6 mm/yr). When using the new uplift rates by Simpson et al. (2012), the SSH trends become larger but stay within the uncertainty intervals of our estimates.

p T S

Hammerfest

Tromsø

Kristiansund

Bergen

Stavanger

Tregde

−0.5 0 0.5 1 −1 trend (mm yr ) Figure 4: Observed trends in the surface pressure effect (in black), and the thermosteric (red) and halosteric (cyan) heights.

11 August 2012

To assess which components are contributing to the observed trend we computed the trends for each individual component (Figure 4). Of the three components air pressure, thermosteric and halosteric effect, it is only the thermosteric effect that has a trend at all locations. Depending on the station, it varies from 0.5-1.0 mm/yr (0.9 mm/yr in Bergen). The pressure effect contributes with a weak trend in Tromsø and Hammerfest while the halosteric effect has a negative trend in Tregde and a positive trend in Tromsø. For comparison, the land uplift rates vary between 1.2-2.7 mm/yr for the stations shown in Figure 4. As already mentioned, there is a difference (residual) between our reconstruction and the observed RSL (see equation 2) and this residual also has a positive trend. This trend is significant, between 1.3-2.3 mm/yr from Tregde to Hammerfest, and it is larger than the combined trend from the three components used in the reconstruction (Figure 4). The conclusion is that processes other than local air pressure, and thermo- and halosteric changes, contribute significantly to the observed positive trends in sea level. Discussion Owing to strong land uplift, the trend in RSL is reduced substantially along the entire Norwegian coast. However, rates of sea level rise appear to be large enough to compensate for vertical land uplift, resulting in positive observed RSL trends along large portions of the coast. This is not unexpected. A similar result has been reported by Rennie and Hansom (2011) for the coast of Scotland where the land uplift is comparable to the uplift in southern and . The positive contribution of the pressure effect in is consistent with the reported decrease of air pressure in the Arctic regions over the past decades (Walsh et al., 1996). The positive trend in the residual sea level indicates that the observed local trends are substantially underestimated by our reconstruction. There are several factors not included into our analysis that may contribute to a rise in observed RSL and these are also discussed in this study. The dominant mode of atmospheric variability over the Nordic Seas is characterized by southwesterly winds that push the water towards the Norwegian coast. By analysing wind data, we find a positive trend for the period 1970-1990. This trend can also be seen in our residual. However, the trend in the residual continues after 1990 while the trend in the wind flattens out. Therefore, it is unlikely that changes in wind forcing have caused the observed long-term rise in RSL. Warming of the deep ocean in the interior of the Nordic Seas can give rise to a redistribution of seawater. When the deep ocean layer expands it will lift the upper ocean layers above the deep ocean but not above the shallower shelf areas along the coasts. This imbalance results in a pressure gradient that drives the water toward the shallow shelf areas. Warming of the deep Nordic Seas has been observed (Østerhus and Gammelsrød, 1999) and is assumed to have taken place since around 1980. For the period 1960-2006, the steric height of the deep ocean increased by 0.4 mm/yr. It is however unclear, how this increase will affect RSL along the coast as it will also induce changes in the ocean circulation. The input of mass from melting land ice results in increased sea levels. However, due to gravitational effects the increase varies from one region to another. By combining published melting rates from the most important sources (Greenland, Antarctica and glaciers) with their so-called fingerprints (the changed global sea level pattern due to the

12 August 2012 effect of ice loss on the Earth's gravity field) we find that melting land ice gives a contribution to the sea level trend on the Norwegian coast for 1972–2008 of 0.7±0.2 mm/yr. This value is comparable to independent gravity measurements (Riva et al., 2010). This effect is comparable to the thermosteric contribution to the trend and explains about one third of the residual trend. Changes in hydrology may also contribute to changing sea levels. It has long been assumed that the loss of water to the sea through irrigation and drainage to rivers is balanced by retaining water on land through the continuous building of dams. However, a recent study indicates that the contribution from land water can be as much as 0.8±0.3 mm/yr in the period 1961-2003 due to the unsustainable use of ground water (Pokhrel et al. 2012). This result is in contrast to gravitational measurements from satellites that show an insignificant global contribution of -0.1±0.3 mm/yr for the period 2003-2009 (Riva et al., 2010). More research is necessary to accurately quantify this component. In addition, the amount of water stored on land varies regionally and with time and changes therefore also the gravitational pull that those water masses exhibit on the water in the ocean. The above mentioned gravity measurements indicate a positive contribution of around 0.4±0.3 mm/yr in our region (Riva et al., 2010). Conclusions We showed that changes in air pressure as well as heat and salt content of sea water account for up to 85% of the observed variability in the relative sea level (RSL) along the Norwegian coast on monthly to interannual time scales. However, these components explain together with land uplift less than half of the observed trends in the absolute sea level. In Bergen, the observed RSL trend is 0.9 mm/yr in the period 1960-2010. In the absence of land uplift the trend would be 2.6 mm/yr (3.4 mm/yr following Simpson et al. 2012). Thermosteric expansion explains 0.9 mm/yr and other factors have to be taken into account to close the trend budget. Some of these factors can be estimated indirectly, like the 0.7±0.2 mm/yr local contribution from the global melting of land ice, or found in the literature, like the 0.8±0.3 mm/yr contribution from changes in hydrology. However, these estimates are subject to large uncertainties and more research is necessary in order to obtain reliable estimates. In addition processes like gravitational effects, wind effects, ocean mass redistribution and changes in circulation influence local sea level and have not yet been measured accurately enough or at all.

13 August 2012

6. Updated estimates of future sea level rise on the Norwegian coast4 H. Drange, J.E.Ø. Nilsen, K. Richter, A. Nesje During the past two decades, global average sea level has risen by about 3 mm/yr. The increase is mainly due to increasing ocean temperatures and melting of land based ice. With the expected rising global temperatures due to greenhouse gas emissions from human activities, the global sea level will continue to rise. Accelerating mass loss from the Greenland and Antarctic ice-sheets can lead to a global sea level rise between one half to one meter by the end of this century, but with large geographic differences. Based on existing literature we estimate a two thirds likelihood of a sea level rise on the Norwegian coast during 100 years of between -10 and 50 cm for Oslo, 25-85 cm for Kristiansand and Stavanger, 20-80 cm for Bergen, -15-40 cm for , and 5-55 cm for Tromsø. Unless global greenhouse emissions are significantly reduced, the sea level rise can be expected to reach the upper half of the given intervals. In a 50 years perspective, sea level rise can be estimated as 30% of the 100-year changes.

Background Global sea level has risen by approximately 120 m since the last glacial maximum approximately 20.000 years ago, mainly due to melting of the great ice sheets of the ice age, but has kept relatively stable during the recent 4000-5000 years (Peltier and Fairbanks, 2006). Despite the strong global rise most of the Norwegian coast has experienced a lowering sea level since the last ice age (Svendsen and Mangerud, 1987). This is because Norway and the rest of Scandianvia have experienced strong land uplift after the Fennoscandian ice sheet melted away and the weight of its mass disappeared (Ekman, 1996; Vestøl, 2006). Thus, changes in the Earth's crust (isostasy) have to be included when describing sea level changes relative to land. The sea level as observed from the shore is called relative sea level. This is the level relevant for populations and coastal infrastructure, and for planning for future sea level rise. Both tide gauge measurements since 1870 and satellite based measurements since 1992 show that the global sea level is rising (Church and White, 2011; Woodworth et al., 2011). The rise is mainly due to increased ocean temperatures leading to (thermal) expansion of the ocean volumes, as well as melting of most glaciers on the Earth (Cazenave and Llovel, 2010). In addition, sea level varies in response to changes in salinity of the ocean (haline contraction). Low salinity, i.e. a fresher ocean, leads to a higher sea level. Averaged over the world oceans, changes in temperature are several times as important for sea level as changes in salinity. In polar regions and along shorelines with fresh river runoff like for Norway, however, salinity can be of comparable importance for sea level as temperature (Richter et al., 2012). Ocean circulation also has an effect on the regional sea level. Due to the Earth's rotation ocean currents will be deflected to the right on the northern hemisphere. If, for instance, the northward flowing North Atlantic Current should slow down, it would contribute to a lowering sea level along the Northern European coasts. An important factor for regional sea level rise are changes in the Earth's gravity field in response to relocation of surface mass, for instance by melting and runoff from glaciers

4 To be cited as Drange, H., J.E.Ø. Nilsen, K. Richter, A. Nesje (2012). Updated estimates of future sea level rise on the Norwegian coast. In Nilsen, J.E.Ø., H. Drange, K. Richter, E. Jansen, A. Nesje. Changes in the past, present, and future sea level on the coast of Norway. NERSC Special Report 89, Bergen, Norway.

14 August 2012 and ice sheets, and changes in the water storage on land (e.g. Mitrovica et al., 2001; Tamisiea and Mitrovica, 2011). Using the Greenland ice sheet as an example, the gravitational effect can be explained as follows (see e.g. Milne et al., 2009): The large mass of ice on Greenland exerts a gravitational pull on its surroundings, and the sea level around Greenland is higher than it would be without this pull. When land ice on Greenland melts two effects come into play: water volumes are added to the ocean and global sea level rises accordingly; at the same time ice, i.e. mass, is removed from Greenland and the gravity field around the ice sheet is reduced. The latter induces a sea level drop in the regions near the ice sheet. The net effect of this is that sea level rise due to a melting Greenland ice sheet mainly takes place in the tropics and the southern hemisphere. Similarly, melting of the Antarctic ice sheet will lead to sea level rise mainly in the tropics and northern hemisphere. This counterintuitive effect was postulated already by the end of the 19th century, but it is only by satellite based gravity measurements during the last decade that this effect on global and regional sea level could be properly quantified (Riva et al. 2010). Observed sea level rise Satellite based measurements of the sea surface height since 1992 shows a global sea level rise around 3 mm/yr (e.g. Leuliette and Willis, 2011). As mentioned, this rise is not uniform over the world oceans, but instead varies following regional changes in ocean temperature, salinity, and circulation, in the gravity field, and in the surface air pressure (Cazenave and Llovel, 2010). Along the Norwegian shores, the average rate of sea level rise during 1891-1990 has been estimated to 1.3 mm/yr (Vestøl, 2006). Further north, in the Arctic Ocean, an average rate of 1.7 mm/yr has been estimated for the period 1962- 1998 (Proshutinsky et al., 2007). The difference in these rates is due to different regions and time periods, as well as uncertainties in the observational basis and methods. Satellite based gravity measurements and the deployment of more than three thousand automated oceanographic floats has made it possible to quantify the different sources' contribution to sea level rise. For the period 2003-2007, the thermal expansion has contributed with about 10%, glaciers with 55%, and the ice sheets of Greenland and Antarctica combined with 40%, as well as a -5% contribution from change in land water storage (Cazenave and Llovel, 2010). Note that the contribution from thermal expansion has been relatively small during the 2000s compared to the 1990s, and that the contribution from glaciers and ice sheets are rapidly increasing. If the thermal expansion increases as expected with global warming and the acceleration of the melting of glaciers and ice sheets continue as during the last decade, global sea level rise can be expected to accelerate in the coming decades. Global sea level in the future According to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC), global sea level rise is expected to range between 18 and 59 cm for the period between 1980-1999 and 2090-2099 (Solomon et al., 2007). It is, however, pointed out by IPCC that this estimate does not include all effects that may lead to sea level rise, such as changes in the flow rates of the ice down from the Greenland and Antarctic ice sheets (Summary for Policymakers, Synthesis Report, IPCC 2007): ”Because understanding of some important effects driving sea level rise is too limited, this report does not assess the likelihood, nor provide a best estimate or an upper bound for sea level rise. Therefore, the upper values of the ranges are not to be considered upper bounds for sea level rise.”

15 August 2012

Several research papers have been published after the Fourth Assessment Report in 2007, see Table 1 for overview. As shown in the table, estimates for global sea level rise span the range from half a meter to two meters. Based on an assessment of this currently available literature, a global sea level rise of less than 50 cm during 100 years is found to be unlikely. Similarly, an upper bound at 110 cm is set for the same period of time. The lower bound at 50 cm corresponds to melting of the Greenland and Antarctic ice sheets at the current rate (National Research Council, 2011), while larger sea level rise can only occur with a significant increase in the melting rates of these two ice sheets (Pfeffer et al., 2008).

Table 1: Overview of estimates of future sea level rise from publications in the time following the IPCC Fourth Assessment report in 2007, based on Nicholls et al. (2011). SRES A1B is an intermediate greenhouse gas emissions scenario. Estimated sea level rise Region Method Source (cm per 100 years) 40-140 Global Empirical projection Rahmstorf (2007) 80-240 Global Past climate reconstruction Rohling et al. (2008) 80-200 Global Upper, physically plausible increase Pfeffer et al. (2008) 50 Global Assessment of lower bound Bahr et al. (2009) 55-110 Global Available literature Vellinga et al. (2008) 56-92 Global Past climate reconstruction Kopp et al. (2009) Vermeer & Rahmstorf 75-190 Global Empirical projection (2009) 72-160 Global Empirical projection Grinsted et al. (2009)

Available literature, National Research Council 50-100 Global Primarily based on SRES A1B (2011) Eastern Global climate model, included 30-80 Katsman et al. (2008) Atlantic gravitational effect Global climate model, including Copen- 40 steric and circulation changes, for Yin et al. (2010) hagen SRES A1B Coast of Global climate models, including 50-80 Slangen et al. (2012 Norway gravitational effect

16 August 2012

Regional sea level in the future Climate models have not yet been run with all known mechanisms affecting global or regional sea level rise. But possible future sea level rise can be estimated by combining the projected sea level change due to changes in ocean temperature, salinity and circulation from climate models, with estimates of contributions from melting glaciers and ice sheets and changes in land water storage, subsequent changes in the gravitational field and the Earth’s crust, and regional land uplift. Estimation of the lower and upper likely values for sea level rise along the Norwegian coast during 100 years, depending on greenhouse gas scenario, can be estimated as follows (the given uncertainties represent one standard deviation): (1) Temperature and salinity: A dozen different climate models used in the IPCC Fourth Assessment Report in 2007 gives an estimate of global sea level rise due to changes in ocean temperatures and salinity of 16±8 cm for scenario SRES B1 (strong reduction in emissions), 21±9 cm for SRES A1B (an intermediate scenario with respect to emissions) and 27±17 cm for SRES A2 (continued high emissions) (Slangen et al. 2012, their Table 4). (2) Total global sea level rise: As mentioned we assume a lower and upper bound on global sea level rise during 100 years of 50 and 110 cm, respectively. If the likelihood for the future sea level rise is following a normal distribution, the lower and upper likely value for global sea level rise may be estimated as 60±5 cm and 90±10 cm. In this case, it is approximately 95% likely that the sea level rise will lie somewhere between the bounds of 50 and 110 cm (two times the standard deviation). (3) Mass contribution from land: The difference between (2) and (1) provides an estimate of the globally averaged contribution from melting of ice on land and changes in land water storage, of between 44±9 cm and 63±20 cm. (4) Present day effect of changes in the gravitational field: For the period 2003- 2009, approximately 40 to 80% of the global mass contribution from land has contributed to sea level rise along the coast of Norway (Riva et al. 2010, their Figure 2; 40% in the area, 80% in Skagerrak). Assuming the future mass loss from land will have a similar geographical distribution and relative contributions, future regional sea level rise due to global mass loss from land can be estimated using the same “fingerprint” pattern. For example, assuming 80% for Oslo leads to a mass contribution to the local sea level in Oslo between 35±7 cm and 50±16 cm (80% of the values in (3)). (5) Future effect of changes in the gravitational field: The geographic distribution of sea level rise caused by melting of glaciers, ice sheets and storage of water on land vary depending on the location and amount of the various sources. The Norwegian coast may receive a slightly negative contribution from Greenland, slightly more than the global mean contribution from Antarctica, and 30-60% of the contribution from glaciers (Riva et al. 2010; Tamisiea and Mitrovica, 2011). The expectation of continued (increased) melting on Greenland indicates a reduced gravitational fingerprint factor compared to that given in (4). When the mass contribution along the Norwegian coast is reduced by 25% (an assumption, for lack of detailed projections of mass contributions from different land sources), one gets approximate fingerprint factors of 30 to 60% along the coast of Norway (75% of the 40 to 80% in (4)). For Oslo this gives a corrected future mass

17 August 2012

contribution to the local sea level between 26±5 cm and 38±12 cm (60% of the values in (3)) (6) Shelf mass loading: For the intermediate emission scenario SRES A1B, an additional 10 cm sea level rise is estimated for the shallow shelf regions off the northwestern European coast and in the Arctic (Meehl et al. 2007, their Figure 10.32; Yin et al. 2010, their Figure 12). Scaling using the global sea level rise estimates in (1) for the three scenarios gives a lower likely value of 8±2 cm (for SRES B1) and an upper likely value of 13±3 cm (for SRES A2) for this additional effect. (7) Land uplift: Land uplift along the Norwegian coast varies from about 10 cm per 100 years along the western coast to about 50 cm per 100 years in the inner Oslofjord and Trondheimsfjord, with an estimated uncertainty of 5 cm (Vestøl, 2006). Estimated land uplift in Oslo is 49±5 cm. (8) Local relative sea level rise: An estimate of the sea level rise relative to land for a 100 years time horizon is then given by the sum of (1), (5) and (6), minus the land uplift from (7). Note that we use the global steric contribution (1), since no significant difference in this contribution, between the Norwegian coast and the global average, has been quantified in the available climate model projections (see e.g. Yin et al. 2010). This gives an estimated future sea level rise for Oslo between 1±11 cm and 29±22 cm. (9) Estimation of likelihood: If the likelihood for future sea level rise is randomly distributed around the lower and upper likely values, a relative sea level rise in Oslo between −10 and +50 cm (rounded to the nearest 5 cm) is 68% likely. For Bergen the values are 20 to 80 cm, and for Tromsø 5 to 55 cm. Values for other coastal cities and communities are given in Figure 5 and Appendix 1. In a recent study by the Norwegian Mapping Authority (Simpson et al., 2012) revised land uplift estimates for the Norwegian coast are presented. These numbers are predominantly somewhat higher than the uplift rates used in our estimation (point 7 above), using data from Vestøl, 2006). The differences between the revised uplift rates and the rates used here are shown in Table 2. The revised values for land uplift are not taken into account in our analysis, but they would for the major part of the Norwegian coast give sea level rise estimates approximately 6 cm lower than our estimates.

Table 2: Differences in values for land uplift during 100 years between those used herein (data from Vestøl, 2006) and the revised numbers from Simpson et al. (2012). Positive differences means higher revised values than those used in our analysis. Oslo +6 cm Stavanger +6 cm Bergen +6 cm Måløy +2 cm Ålesund +8 cm Kristiansund +9 cm Rørvik +6 cm Bodø +6 cm Narvik +1 cm Harstad +6 cm Tromsø +1 cm Hammerfest +5 cm Honningsvåg −6 cm

For a time perspective of 50 years, the sea level increase can be estimated as about 30% of the given values for 100 years (based on Rahmstorf, 2007, and Drange et al., 2007).

18 August 2012

Figure 5: Estimates of sea level rise in cm relative to the shore in a 100 years perspective. The limits are based on one standard deviation, indicating 68% likelihood that the sea will rise to a level between the given values. Values for 95% likelihood are given in Table 4.

The large span in the estimates of future sea level rise – as illustrated in Figure 5 and tabulated in Appendix 1 – is mainly due to the uncertainties in the amount of melting of the land based ice (i.e. Greenland and Antarctica) during this century as a result of future greenhouse gas emissions, and how the Earth’s gravity field will change as a result of this. Note that the shelf mass loading described in (6) and melting in the Antarctic, will counter some of the negative contribution from the gravity effect of melting on Greenland described in (5). Furthermore, any gravity effects (“self attraction”) by accumulated waters on continental shelves are still not quantified by gravity models, and hence not included in the above analysis. The most complete analysis of future sea level to date is likely the one by Slangen et al. (2012). They use 12 global climate models (the Bergen Climate Model included) and the above-mentioned effects are included in the study. The projection in Slangen et al. (2012) assumes a mass exchange contribution from land (glaciers and ice sheets) between 21 and 28 cm during 100 years, where Antarctica is assumed to contribute approximately 1 cm and Greenland 6-8 cm to global sea level rise during 100 years. These contributions are relatively small, and notably less than the mass contribution we estimate in (3) above (44 to 63 cm). The methods of Slangen et al. (2012) are transferred to the Norwegian coast by Simpson et al. (2012), yielding an estimated sea level rise for the period 2090-2099 relative to 1980-1999 of between −20 and +30 cm. Taking into account that our numbers for mass contribution are 23 to 35 cm larger than those used by Slangen/Simpson (21-28 cm compared to 44-63 cm), there is a general agreement between Simpson et al. (2012) and our results. Differences are, in other words, mainly due to projections of mass contributions from glaciers and ice sheets. Furthermore, Simpson et al. (2012) also estimate the possible sea level rise along the Norwegian coast using a scenario with (very) large mass contribution from glaciers and ice sheets. This scenario yields estimates of sea level rise along the coast between 70 and 130 cm for the period 2090-2099 relative to 1980-1999. Even though this scenario cannot be excluded, there is currently no observational support for such a large and fast mass exchange to the ocean. The values from the analysis herein are somewhat lower than the previous estimates of sea level rise by Vasskog et al. (2009) and Hanssen-Bauer et al. (2010). The main difference is

19 August 2012 due to the inclusion of gravity effects in the current analysis (but mentioned in the other two). Furtermore, previous estimates were given as upper and lower bounds, thus comparison is more relevant using a two standard deviation limit on the new likely values (giving 95% likelihood that sea level rise will fall between the values). Such a comparison is provided for the major coastal cities in Table 3. The lower boundaries exhibit the largest differences, with the present estimates being 50 cm lower than the previous estimates. There are smaller differences at the upper bounds, with present estimates 5-25 cm lower than the previous estimates.

Table 3: Comparisons of the lower and upper bounds in previous estimates of sea level rise (Vasskog et al., 2009; Hanssen-Bauer et al., 2010) and the present estimates in this analysis. Boundaries for the latter are expressed using two standard deviations. Unit is cm, and estimates are for a change over 100 years. Previous estimates Present estimates (lower and upper (95% likelihood) bounds) Min. Max. Min. Max.

Tromsø +45 +100 −5 +75 Trondheim +20 +75 −25 +60 Bergen +55 +110 +10 +100 Stavanger +60 +115 +15 +110 Oslo +20 +75 −20 +70

More numerous and accurate observations and rapidly increasing theoretical knowledge about current and possible future sea level rise will necessitate regularly updated projections of global and local sea level rise. In this respect it is worth noting that global and regional sea level rise will be given a separate chapter in the next assessment report from IPCC, coming in fall 2013. In the forthcoming IPCC-report, the uncertainties will be discussed and, as far as possible, quantified. The knowledge and understanding that will be presented in the next IPCC-report will thus be a logical starting point for revised estimates of future sea level rise along the Norwegian coast. For an updated assessment of how some of the major cities in the world adapt to storm surges and rising sea level, we refer to Becker et al. (2012), Hallegatte et al. (2011), and Hanson et al. (2011). For instance Amsterdam, Rotterdam, London, Tokyo, and Shanghai are protected from more than a one-in-a-thousand years storm surge event, while New York has protection corresponding to a one-in-a-hundred years event (Hanson et al., 2011).

20 August 2012

7. Dissemination

Richter, K. (2010 and 2012), Sea level - past, present and future. International youth camp (ICE), Husavik, Iceland and Ryvar, Norway, June 2010, 2012 (oral) Richter, K., J.E.Ø. Nilsen and H. Drange (2012), Contributions to sea level variability along the Norwegian coast for 1960-2010, CRES workshop on sea level rise, Copenhagen, Denmark, 23.05.2012 (oral). Nilsen, J.E.Ø. (2012). Sea level change and ice sheet dynamics. ECRA pilot workshop on regional sea level change, Utrecht, Netherlands, 15.03.2012. Richter, K., J.E.Ø. Nilsen and H. Drange (2011), Contributions to sea level variability along the Norwegian during the past 50 years, Centre for Climate Dynamics opening, Bergen, Norway, 02.12.2011 (oral). Nilsen, J.E.Ø., K. Richter and H. Drange (2011). Changes in past, present and future sea level, focusing on the Norwegian west coast. Lecture for Danish Institute for Study Abroad (DIS), Bergen, 26.06.2011 (oral). Richter, K., J.E.Ø. Nilsen and H. Drange (2011). Sea level variability along the Norwegian coast. EGU General Assembly, Vienna, Austria, 03.04.2011. Nilsen, J.E.Ø., K. Richter and H. Drange (2010)."Changes in the past, present, and future sea level on the coast of Norway". "Floods and Sea Level Rise in the Cities of the Future", TEKNA conference, Stavanger, 26.04.2010 (oral). Nilsen, J.E.Ø., K. Richter and H. Drange (2010). Changes in past, present and future sea level, focusing on the Norwegian west coast. MARE Climate Forum, Bergen, 25.05.2010 (oral). Nilsen, J.E.Ø., K. Richter and H. Drange (2010). Changes in past, present and future sea level, focusing on the Norwegian west coast. Norwegian Geophysical Society Symposium, Geilo, 17.09.2010 (oral).

As well as numerous talks and newspaper articles by H. Drange, approximately 30-50 per year.

21 August 2012

Bibliography Bahr, D. B. et al. (2009), Sea-level rise from glaciers and ice caps: A lower bound, Geophys. Res. Lett., doi:10.1029/2008GL036309 Bard, E., B. Hamelin, and R. G. Fairbanks, 1990: U-Th ages obtained by mass spectrometry in corals from Barbados: sea level during the past 130,000 years. Nature, 346, 456-458. Becker, A. et al. (2012), Climate change impacts on international seaports: knowledge, perceptions, and planning efforts among port administrators, Clim. Change, doi:10.1007/s10584-011-0043-7 Blanchon, P., A. Eisenhauer, J. Fietzke, and V. Liebetrau, 2009: Rapid sea level rise and reef back-stepping at the close of the last interglacial highstand. Nature, 458, 881-884. Brigham-Grette, J., and L. Carter, 1992: Pliocene marine transgressions of northern Alaska: circumarctic correlations and paleoclimatic interpretations. Arctic, 45, 74-89. Cazenave, A. og W. Llovel (2010), Contemporary sea level rise. Annu. Rev. Mar. Sci. 2, 145–173. doi:10.1146/annurev-marine-120308-081105 Church, J. A. og N. J. White (2011), Sea-level rise from the late 19th to the early 21st century, Surv. Geophys., DOI 10.1007/s10712-011-9119-1 Drange, H., B. Marzeion, A. Nesje, A. Sorteberg (2007), Opptil en meter havstigning langs Norskekysten innen år 2100, Cicerone, 2, 29-31 Ekman M (1996), A consistent map of the postglacial uplift of Fenno- scandia. Terra Nova 8,158–165 Grinsted, A. et al. (2009), Reconstructing sea level from paleo and projected temperatures 200 to 2100 AD, Clim. Dyn., doi:10.1007/s00382- 008-0507-2 Hallegatte, S. et al. (2011), Assessing climate change impacts, sea level rise and storm surge risk in port cities: a case study on Copenhagen, Clim. Change, doi:10.1007/s10584-010-9978-3 Hanebuth, T., K. Stattegger, and P. M. Grootes, 2000: Rapid Flooding of the Sunda Shelf: A Late-Glacial Sea level Record. Science, 288, 1033-1035. Hanson, S. et al. (2011), A global ranking of port cities with high exposure to climate extremes, Clim. Change, doi: 10.1007/s10584-010-9977-4 Hanssen-Bauer, I. et al. (2009), Klima i Norge 2100. Bakgrunnsmateriale til NOU Klimatilplassing, Norsk klimasenter, september 2009, Oslo Katsman, C. A. et al. (2008), Climate scenarios of sea level rise for the northeast Atlantic Ocean: a study including the effects of ocean dynamics and gravity changes induced by ice melt, Climatic Change, 91, 351–374 Kopp, R. et al. (2009), Probabilistic assessment of sea level during the last interglacial stage. Nature, doi:10.1038/nature08686 Leuliette, E. W. og J. K. Willis (2011), Balancing the sea level budget. Oceanogr. 24,122–129, doi:10.5670/oceanog.2011.32 Lisiecki, L., and M. Raymo, 2005: A Pliocene-Pleistocene stack of 57 globally distributed benthic !18O records. Paleoceanography, 20, PA1003. Lohne, Ø. S., S. Bondevik, J. Mangerud, J. I. Svendsen (2007). Sea-level fluctuations imply that the Younger Dryas ice-sheet expansion in western Norway commenced during the Allerød. Quaternary Science Reviews 26, doi: 10.1016/j.quascirev.2007.04.008. McKay, N. P., J. T. Overpeck, and B. L. Otto-Bliesner, 2011: The role of ocean thermal expansion in Last Interglacial sea level rise. Geophysical Research Letters, 38, L14605. Meehl, G., et al. (2007), Global climate projections, in Climate Change 2007: The Physical Science Basis, edited by S. Solomon et al., pp. 847–940, Cambridge Univ. Press, Cambridge, U. K. Miller, K., et al., in press: The high tide of the warm Pliocene: Implications of global sea level for Antarctic deglaciation. Geology. Milne, G. A. et al. (2009), Identifying the causes of sea-level change, Nature Geosci., doi: 10.1038/ngeo544 Mitrovica, J. X. et al. (2001), Recent mass balance of polar ice sheets inferred from patterns of global sea- level change. Nature 409, 1,026–1,029 Naish, T., et al., 2009: Obliquity-paced Pliocene West Antarctic ice sheet oscillations. Nature, 458, 322- U384. National Research Council (2011), Climate Stabilization Targets: Emissions, Concentrations, and Impacts over Decades to Millennia. Washington, DC: National Academies Press (ISBN 978-0-309-15176-4) Nicholls, R. J. et al. (2011), Sea-level rise and its possible impacts given a ‘beyond 4°C world’ in the twenty- first century, Phil. Trans. R. Soc. A, doi: 10.1098/rsta.2010.0291. Peltier, W. R. (2004), Global glacial isostasy and the surface of the ice-age Earth: The ICE-5G(VM2) model and GRACE, Annu. Rev. Earth Planet. Sci., 32, 111–149. Peltier, W. R. og R. G. Fairbanks (2006), Global glacial ice volume and Last Glacial Maximum duration from an extended Barbados sea level record. Quaternary Sci. Rev. 25, 3322–3337

22 August 2012

Pfeffer, W. et al. (2008), Kinematic constraints on glacier contributions to 21st-century sea-level rise. Science, doi:10.1126/science.1159099 Pokhrel Y. N., N. Hanasaki, P. J-F. Yeh, T. J. Yamada, S. Kanae, T. Oki (2012). Model estimates of sea-level change due to anthropogenic impacts on terrestrial water storage. Nature Geoscience, 5, doi: 10.1038/NGEO1476. Pollard, D., and R. M. DeConto, 2009: Modelling West Antarctic ice sheet growth and collapse through the past five million years. Nature, 458, 329-332. Proshutinsky, A. et al. (2007), Sea level variability in the Arctic Ocean from AOMIP models, J. Geophys. Res., doi:10.1029/2006JC003916. Rahmstorf, S. (2007), A semi-empirical approach to projecting future sea-level rise. Science 315, 368-370 Rennie, A., and J. Hansom (2011), Sea level trend reversal: Land uplift outpaced by sea level rise on Scotland’s coast, Geomorphology, 125, 193–202. Richter, K., J.E.Ø. Nilsen, H. Drange (2012). Contributions to sea level variability along the Norwegian coast for 1960-2010. J. Geophys. Res., doi:10.1029/2009JC007826. Riva, R. E. M. et al. (2010), Sea level fingerprint of continental water and ice mass change from GRACE, Geophys. Res. Lett., 37, L19605, doi:10.1029/ 2010GL044770 Rohling, E. et al. (2008), High rates of sea-level rise during the last interglacial period. Nat. Geosci., doi:10.1038/ngeo.2007.28 Rohling, E. J., K. Grant, M. Bolshaw, A. P. Roberts, M. Siddall, C. Hemleben, and M. Kucera, 2009: Antarctic temperature and global sea level closely coupled over the past five glacial cycles. Nature Geoscience, 2, 500- 504. Simpson, M., Breili, K., Kierulf, H. P., Lysaker, D., Ouassou, M., and Haug, E. (2012). Estimates of Future Sea-Level Changes for Norway. Technical Report of the Norwegian Mapping Authority [download] Slangen, A. B. A. m.fl (2012), Towards regional projections of twenty-first century sea-level change based on IPCC SRES scenarios, Clim. Dyn., doi:10.1007/s00382-011-1057-6 Solomon, S. et al. (2007): Technical Summary. In: Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [Solomon, S. et al. (forf.)]. Cambridge University Press, Cambridge, United Kingdom og New York, NY, USA Svendsen, J. I. og J. Mangerud (1987), Late Weichselian and Holocene Sea-Level History for a Cross- Section of Western Norway. J. Quaternary Sci. 2, 113 - 132. Tamisiea, M. E. og J. X. Mitrovica (2011), The moving boundaries of sea level change: Understanding the origins of geographic variability. Oceanogr. 24, 24–39, doi:10.5670/ oceanog.2011.25 Vasskog, K. et al. (2009), Havnivåstigning. Estimater av framtidig havnivåstigning i norske kystkommuner. Det nasjonale klimatilpasningssekretariatet ved Direktoratet for samfunnssikkerhet og beredskap, september 2009, Tønsberg Vellinga, P. et al. (2008), Exploring high-end climate change scenarios for flood protection of The Netherlands. International Scientific Assessment carried out at request of the Delta Committee. Scientific report WR-2009-05. KNMI, Alterra, The Netherlands (http://www.knmi.nl/ bibliotheek/knmipubWR/WR2009-05.pdf) Vermeer, M. og S. Rahmstorf (2009), Global sea level linked to global temperature. Proc. Natl Acad. Sci., doi:10.1073/pnas.0907765106 Vestøl, O. (2006), Determination of postglacial land uplift in Fennoscandia from levelling, tide-gauges and continuous GPS stations using least squares collocation. J. Geodesy 80: 248-258 Walsh, J. E., W. L. Chapman, and T. L. Shy (1996), Recent decrease of sea level pressure in the central Arctic, J. Clim., 9, 480–486. Woodworth, P. L. et al. (2011), Nineteenth and twentieth century changes in sea level. Oceanogr. 24, 80–93, doi:10.5670/oceanog.2011.29 Yin, J. et al. (2010), Spatial variability of sea level rise in twenty-first century projections, J. Climate, doi:10.1175/2010JCLI3533.1 Østerhus, S., and T. Gammelsrød (1999), The abyss of the Nordic seas is warming, J. Clim., 12, 3297–3304.

23 August 2012

Appendix Appendix 1: Estimates of future sea level rise for the Norwegian coastal municipalities Table 4: Overview of estimated sea level rise during 100 years for the Norwegian coastal municipalities, rounded to nearest 5 cm. The table is an extension of Vasskog et al. (2009) and Hanssen-Bauer et al. (2010), but with estimates based on the analysis in the points (1)-(9) in Chapter 6. Light grey shaded columns contain lower and upper change in relative sea level with 68% estimated likelihood (one standard deviation from the calculated limit), while dark grey columns contain lower and upper change with 95% estimated likelihood (two standard deviations from the calculated limit). The table also show the land uplift estimates used (from Vestøl 2006) and gravitational fingerprint factor used (based on Figure 2 in Riva et al., 2010; see point (5) in Chapter 6). Based on Rahmstorf (2007) and Drange et al. (2007), the 50 years sea level rise can be estimated at approximately 30% of the tabulated values. The revised land uplift rates from Simpson et al. (2012) are not included in the analysis, but an approximate adjustment towards Simpson et al. can be done by deducting 6 cm from the tabulated values. Gravitational 68% likelihood 95% likelihood Fingerprint Lower Upper Lower Upper Land uplift Factor Limit limit limit limit Finnmark (cm) (%) (cm) (cm) (cm) (cm) Municipality Observation point Sør-Varanger Kirkenes 30 75 5 65 -5 85 Nesseby Nesseby 25 75 10 70 0 90 Vadsø Vadsø 26 75 10 70 0 90 Vardø Vardø 22 75 15 75 5 95 Båtsfjord Båtsfjord 23 75 15 75 5 95 Berlevåg Berlevåg 23 75 15 75 5 95 Tana Smalfjord 24 75 15 70 5 95 Gamvik Gamvik 20 75 20 75 5 100 Lebesby Lebesby 25 75 10 70 0 90 Nordkapp Honningsvåg 22 75 15 75 5 95 Porsanger Lakselv 31 75 5 65 -5 85 Måsøy Havøysund 23 75 15 75 5 95 Kvalsund Kvalsund 26 75 10 70 0 90 Hammerfest Hammerfest 25 75 15 70 0 95 Hasvik Breivikbotn 26 75 10 70 0 90 Alta Alta 31 75 5 65 -5 85 Loppa Øksfjord 31 75 5 65 -5 85

Troms Municipality Observation point Land uplift Gravity Min 68% Max 68% Min 95% Max 95% Kvænangen Burfjord 32 50 0 50 -10 70 Nordreisa Sørkjosen 34 50 -5 50 -15 70 Skjervøy Skjervøy 33 50 0 50 -15 70 Kåfjord Olderdalen 36 50 -5 45 -15 65 Storfjord Skibotn 35 50 -5 50 -15 70 Lyngen Lyngseidet 34 50 -5 50 -15 70 Karlsøy Karlsøy 29 50 0 55 -10 75 Tromsø Tromsø (Breidvika) 27 50 5 55 -5 75 Sommarøy 26 50 5 55 -5 75 Balsfjord Storsteinnes 34 50 -5 50 -15 70 Mortenhals 32 50 0 50 -10 70 Målselv Målsnes 32 50 0 50 -10 70 Lenvik Finnsnes 32 50 0 50 -10 70 Berg Skaland 27 50 5 55 -5 75 Torsken Gryllefjord 26 50 5 55 -5 75 Tranøy Vangsvik 33 50 0 50 -15 70 Sørreisa Sørreisa 34 50 -5 50 -15 70 Dyrøy Brøstadbotn 33 50 -5 50 -15 70 Salangen Sjøvegan 40 50 -10 45 -20 60 Lavangen Tennevoll 41 50 -10 40 -20 60 Gratangen Årstein 42 50 -10 40 -20 60 Ibestad Hamnvik 34 50 -5 50 -15 70 Skånland** Evenskjer 34 50 -5 50 -15 70 Bjarkøy Nergårshamn 25 50 5 60 -5 75 Harstad Harstad 27 50 5 55 -5 75 Kvæfjord Borkenes 29 50 0 55 -10 75

24 August 2012

Nordland Municipality Observation point Land uplift Gravity Min 68% Max 68% Min 95% Max 95% Andøy 24 40 5 55 -5 75 Øksnes Myre 25 40 0 55 -10 70 Sortland 26 40 0 50 -10 70 Bø Straume 24 40 5 55 -5 75 25 40 0 55 -10 70 Vågan** Svolvær (Sør) 26 40 0 50 -10 70 Laukvika (Nord) 25 40 5 55 -5 70 Vestvågøy** (Sør) 24 40 5 55 -5 75 Eggum (Nord) 23 40 5 55 -5 75 ** Ramberg (Nord) 24 40 5 55 -5 75 Nusfjord (Sør) 24 40 5 55 -5 75 ** Reine (Sør) 23 40 5 55 -5 75 (Nord) 22 40 5 55 -5 75 Værøy** Sørland (Sør) 25 40 5 55 -5 70 Flyplass (Nord) 25 40 5 55 -5 70 Røst Røstlandet 21 40 5 55 -5 75 Lødingen Lødingen 33 40 -5 45 -15 65 ** Nedre Fjeldal (Nord) 34 40 -5 45 -15 65 Ramsund (Sør) 36 40 -10 40 -20 60 Bogen 41 40 -15 35 -25 55 Narvik Narvik 44 40 -15 35 -25 50 Ballangen 40 40 -10 40 -20 55 Kjøpsvik 38 40 -10 40 -20 60 Hamarøy Oppeid 35 40 -10 40 -20 60 Leinesfjorden 36 40 -10 40 -20 60 Sørfold Straumen 46 40 -20 30 -30 50 Bodø*** Bodø 36 40 -10 40 -20 60 Skjerstadfjorden 40 40 -15 40 -25 55 Fauske 43 40 -15 35 -25 55 Rognan 47 40 -20 30 -30 50 * Misvær 40 40 -15 40 -25 55 ** Moldjord (Leirvika) 43 40 -15 35 -25 55 Gildeskål Inndyr 40 40 -15 35 -25 55 Meløy Ørnes 44 40 -15 35 -25 55 Rødøy Våga 40 40 -15 40 -25 55 Rana 53 40 -25 25 -35 45 Træna Selvær 31 40 -5 45 -15 65 Lurøy Lurøy 40 40 -15 40 -25 55 Nesna 44 40 -15 35 -25 55 Leland 44 40 -15 35 -25 55 Bjerka 51 40 -25 25 -35 45 Mosjøen 46 40 -20 30 -30 50 Dønna Solfjellsjøen 40 40 -15 40 -25 55 Herøy Silvalen 45 40 -15 35 -25 50 Sandnessjøen 44 40 -15 35 -25 50 Vega Holand 41 40 -15 35 -25 55 Forvik 46 40 -20 30 -30 50 Brønnøy Brønnøysund 46 40 -20 30 -30 50 Sømna Vik (Sørvika) 46 40 -20 30 -30 50 Terråk 51 40 -25 25 -35 45 * Municipalities with two observation points ** Skjerstad municipality merged with Bodø in 2005

*** See desription in Section 2.3 of Vasskog et al. (2009)

25 August 2012

Nord-Trøndelag Municipality Observation point Land uplift Gravity Min 68% Max 68% Min 95% Max 95% Leka Sør-Gutvika 47 50 -15 35 -25 55 Nærøy 47 50 -15 35 -25 55 Høylandet Kongsmoen 53 50 -20 30 -35 50 Rørvik 43 50 -10 40 -25 60 Salsnes 46 50 -15 35 -25 55 Namsos 47 50 -15 35 -25 55 Lauvsnes 42 50 -10 40 -20 60 Sjøåsen 45 50 -15 40 -25 60 47 50 -15 35 -25 55 Steinkjer 51 50 -20 30 -30 50 Inderøy Straumen 51 50 -20 30 -30 50 Leksvik 48 50 -20 35 -30 55 Saltvikhamna 50 50 -20 35 -30 50 Verdal 51 50 -20 30 -30 50 Levanger 53 50 -20 30 -35 50 Sørgrenda 50 50 -20 35 -30 55 Stjørdal Stjørdalshalsen 52 50 -20 30 -30 50

Sør-Trøndelag Municipality Observation point Land uplift Gravity Min 68% Max 68% Min 95% Max 95% Osen 40 60 -5 50 -15 70 Roan Roan 39 60 -5 50 -15 70 Åfjord Årnes 41 60 -5 50 -20 70 * Botngård 38 60 -5 50 -15 70 Høybakken 38 60 -5 50 -15 70 Frøya Sistranda 29 60 5 60 -5 80 Ørland* 38 60 -5 50 -15 70 Uthaug 36 60 -5 50 -15 70 Rissa Rissa 41 60 -5 50 -20 70 Fillan 31 60 5 60 -10 80 * Krogstadøra 36 60 -5 50 -15 70 Futstranda 33 60 0 55 -10 75 * Lensvik 40 60 -5 50 -15 70 Stavøysundet 36 60 -5 50 -15 70 Kyrksæterøra 32 60 0 55 -10 75 42 60 -10 45 -20 65 Børsa 45 60 -10 45 -20 65 Gran 47 60 -15 40 -25 60 Trondheim Trondheim 48 60 -15 40 -25 60 Hommelvik 54 60 -20 35 -30 55 *Municipalities with two observation points

26 August 2012

Møre og Romsdal Municipality Observation point Land uplift Gravity Min 68% Max 68% Min 95% Max 95% Smøla Hopen 27 60 5 60 -5 80 Aure Aure 29 60 5 60 -5 80 Vågland 30 60 5 60 -5 80 Surnadal Surnadalsøra 33 60 0 55 -10 75 Kristiansund Kristiansund 26 60 10 65 -5 85 Rensvik 27 60 5 60 -5 80 Tingvoll Tingvoll 30 60 5 60 -5 80 Sunndal Sunndalsøra 33 60 0 55 -10 75 Averøy Kårvåg 25 60 10 65 0 85 Gjemnes Batnfjordsøra 25 60 10 65 -5 85 Nesset Eidsvåg 29 60 5 60 -5 80 Eide Eide 25 60 10 65 0 85 Fræna Elnesvågen 23 60 10 65 0 85 Molde Molde 23 60 10 65 0 85 Rauma Åndalsnes 27 60 5 60 -5 80 Aukra Aukrasanden 23 60 10 65 0 85 Sandøy Steinshamn 22 60 10 65 0 85 Midsund Midsund 21 60 10 65 0 85 Vestnes Helland 24 60 10 65 0 85 Haram Brattvåg 21 60 15 70 0 90 Skodje Skodje 21 60 15 70 0 90 Ørskog Sjøholt 22 60 10 65 0 85 Stordal Stordal 22 60 10 65 0 85 Norddal Sylte 26 60 10 65 -5 85 Giske Valderhaugstranda 19 60 15 70 5 90 Ålesund Ålesund 19 60 15 70 5 90 Sykkylven Aure 20 60 15 70 5 90 Stranda Stranda 23 60 10 65 0 85 Ulstein Ulsteinvik 18 60 15 70 5 90 Hareid Hareid 18 60 15 70 5 90 Sula Langevågen 18 60 15 70 5 90 Ørsta Ørsta 19 60 15 70 5 90 Herøy Fosnavåg 18 60 15 70 5 90 Volda Volda 18 60 15 70 5 90 Sande Larsnes 18 60 15 70 5 90 Vanylven Fiskå 19 60 15 70 5 90

Sogn og Fjordane Municipality Observation point Land uplift Gravity Min 68% Max 68% Min 95% Max 95% Selje Selje 19 60 15 70 5 90 Vågsøy Måløy 19 60 15 70 5 90 Eid Nordfjordeid 19 60 15 70 5 90 Stryn Stryn 21 60 10 70 0 90 Bremanger Svelgen 20 60 15 70 0 90 Gloppen Sandane 21 60 10 65 0 85 Flora Florø 20 60 15 70 5 90 Naustdal Naustdal 23 60 10 65 0 85 Luster Gaupne 26 60 5 60 -5 80 Askvoll Askvoll 20 60 15 70 0 90 Førde Førde 24 60 10 65 0 85 Fjaler Dale 21 60 15 70 0 90 Gaular Bygstad 23 60 10 65 0 85 Balestrand Balestrand 23 60 10 65 0 85 Leikanger Leikanger 24 60 10 65 0 85 Sogndal Sogndal 25 60 10 65 0 85 Årdal Årdalstangen 28 60 5 60 -5 80 Hardbakke 19 60 15 70 5 90 Hyllestad Hyllestad 20 60 15 70 0 90 Høyanger Høyanger 23 60 10 65 0 85 Vik Vik 23 60 10 65 0 85 Aurland Aurlandsvangen 24 60 10 65 0 85 Lærdal Lærdalsøyri 27 60 5 60 -5 80 Eivindvik 18 60 15 70 5 90

27 August 2012

Hordaland Municipality Observation point Land uplift Gravity Min 68% Max 68% Min 95% Max 95% Fedje 18 75 20 80 10 100 Fonnes 18 75 20 80 10 100 Solheim 19 75 20 80 10 100 Mo 19 75 20 80 10 100 Radøy 18 75 20 80 10 100 Lindås Knarvik 17 75 20 80 10 100 Vaksdal Vaksdal 18 75 20 80 10 100 Voss* Bolstadøyri 18 75 20 80 10 100 Øygarden Tjeldstø 17 75 20 80 10 100 Meland Frekhaug 17 75 20 80 10 100 Osterøy Lonevåg 17 75 20 80 10 100 Fjell Straume 17 75 20 80 10 100 Askøy Kleppestø 17 75 20 80 10 100 Bergen Bergen 17 75 20 80 10 100 Samnanger Tysse 17 75 20 80 10 100 Kvam Norheimsund 18 75 20 80 10 100 Granvin Eide 21 75 15 75 5 95 Ulvik Ulvik 23 75 15 75 5 95 Sund Tælavåg 17 75 20 80 10 100 Austevoll Storebø 16 75 20 80 10 100 Os Osøyro 16 75 20 80 10 100 Fusa Eikelandsosen 16 75 20 80 10 100 Jondal Jondal 17 75 20 80 10 100 Ullensvang Kinsarvik 20 75 15 75 5 95 Eidfjord Eidfjord 23 75 15 75 5 95 Tysnes Uggdalseidet 15 75 20 80 10 100 Bømlo Svortland 14 75 25 85 15 105 Fitjar Fitjar 14 75 25 80 10 105 Stord Leirvik 14 75 25 85 15 105 Kvinnherad Rosendal 16 75 20 80 10 100 Odda Odda 17 75 20 80 10 100 Sveio Mølstrevåg 12 75 25 85 15 105 Etne Etne 15 75 25 80 10 105 * See description in Section 2.3 of Vasskog et al. (2009).

Rogaland Municipality Observation point Land uplift Gravity Min 68% Max 68% Min 95% Max 95% Haugesund Haugesund 10 80 30 90 20 110 Vindafjord Ølen 14 80 25 85 15 105 Sandeid 14 80 25 85 15 105 Sauda Sauda 16 80 25 85 10 105 Utsira 10 80 30 90 20 110 Karmøy Kopervik 10 80 30 90 20 110 Tysvær Hervik 12 80 25 90 15 110 Suldal Sand 16 80 25 85 10 105 Bokn Føresvik 11 80 30 90 15 110 Finnøy Judaberg 13 80 25 85 15 110 Hjelmeland Hjelmeland 14 80 25 85 15 105 Kvitsøy Ystabøhamn 10 80 30 90 20 110 Rennesøy Vikevåg 11 80 30 90 15 110 Randaberg Bøvika 11 80 30 90 15 110 Stavanger Stavanger 12 80 30 90 15 110 Strand Jørpeland 13 80 25 85 15 110 Sola Solavika 11 80 30 90 15 110 Sandnes Sandnes 12 80 25 85 15 110 Forsand Forsand 13 80 25 85 15 110 Klepp Revtangen 11 80 30 90 15 110 Gjesdal Frafjord 11 80 30 90 15 110 Hå Sirevåg 10 80 30 90 20 110 Eigersund Eigersund 9 80 30 90 20 110 Sokndal Sogndalsstranda 9 80 30 90 20 110

28 August 2012

Vest-Agder Municipality Observation point Land uplift Gravity Min 68% Max 68% Min 95% Max 95% Flekkefjord Flekkefjord 9 80 30 90 20 110 Kvinesdal Øye 9 80 30 90 20 110 Farsund Farsund 10 80 30 90 20 110 Lyngdal Lyngdal 11 80 30 90 15 110 Lindesnes Åvik 12 80 30 90 15 110 Mandal Mandal 13 80 25 85 15 110 Søgne Høllen 14 80 25 85 15 105 Kristiansand Kristiansand 16 80 25 85 10 105

Aust-Agder Municipality Observation point Land uplift Gravity Min 68% Max 68% Min 95% Max 95% Lillesand Lillesand 20 80 20 80 10 100 Grimstad Grimstad 22 80 15 75 5 100 Arendal Arendal 24 80 15 75 5 95 Tvedestrand Tvedestrand 25 80 15 75 5 95 Risør Risør 26 80 15 75 0 95

Telemark Municipality Observation point Land uplift Gravity Min 68% Max 68% Min 95% Max 95% Kragerø Kragerø 26 80 15 75 0 95 Bamble Langesund 30 80 10 70 0 90 Porsgrunn Porsgrunn 32 80 5 65 -5 90

Vestfold Municipality Observation point Land uplift Gravity Min 68% Max 68% Min 95% Max 95% Larvik Larvik 32 80 10 70 -5 90 Sandefjord 34 80 5 65 -5 85 Tjøme Verdens Ende 34 80 5 65 -5 85 Melsomvik 36 80 5 65 -10 85 Nøtterøy Årøysund 37 80 5 65 -10 85 Tønsberg Tønsberg 36 80 5 65 -10 85 Horten 39 80 0 60 -10 80 Re Mulodden 40 80 0 60 -10 80 Holmestrand Holmestrand 40 80 0 60 -10 80 Sande Selvik 42 80 -5 60 -15 80 Svelvik Svelvik 42 80 -5 60 -15 80

29 August 2012

Buskerud Municipality Observation point Land uplift Gravity Min 68% Max 68% Min 95% Max 95% Drammen Drammen (Tangen) 45 80 -5 55 -15 75 Lier Linnesstranda 45 80 -5 55 -15 75 Røyken Nærsnes 45 80 -5 55 -15 75 Hurum Tofte 43 80 -5 55 -15 80

Oslo Municipality Observation point Land uplift Gravity Min 68% Max 68% Min 95% Max 95% Oslo Oslo 49 80 -10 50 -20 70

Akershus Municipality Observation point Land uplift Gravity Min 68% Max 68% Min 95% Max 95% Asker Konglungen 47 80 -5 55 -20 75 Bærum Sandvika 48 80 -10 50 -20 75 Nesodden Nesoddtangen 48 80 -10 50 -20 75 Oppegård Svartskog 46 80 -5 55 -20 75 Frogn Drøbak 44 80 -5 55 -15 75 Ås Nesset 46 80 -5 55 -20 75 Vestby Son 39 80 0 60 -10 80

Østfold Municipality Observation point Land uplift Gravity Min 68% Max 68% Min 95% Max 95% Moss Moss 39 80 0 60 -10 80 Rygge Rørvik 38 80 0 60 -10 85 Råde Saltnes 38 80 0 60 -10 85 Fredrikstad* Fredrikstad 38 80 0 60 -10 85 Sarpsborg* Høysand 41 80 0 60 -15 80 Hvaler Skjærhollen 37 80 0 60 -10 85 Halden Halden 42 80 -5 55 -15 80 * See description in Section 2.3 of Vasskog et al. (2009)

30 August 2012

Appendix 2: Peer review publication on sea level change during the past 50 years

Richter, K., J.E.Ø. Nilsen, H. Drange (2012) Contributions to sea level variability along the Norwegian coast for 1960-2010 J. Geophys. Res., 117, doi:10.1029/2009JC007826

31 August 2012

32 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 117, C05038, doi:10.1029/2011JC007826, 2012

Contributions to sea level variability along the Norwegian coast for 1960–2010 K. Richter,1,2 J. E. Ø. Nilsen,2,3 and H. Drange1,2,4 Received 13 December 2011; revised 20 April 2012; accepted 25 April 2012; published 26 May 2012.

[1] Global sea level has been rising by about 20 cm during the last century and is expected to continue to rise in the 21st century. The rise and variability is not spatially uniform. To be able to project local changes in relative sea level (RSL), it is important to identify the processes that govern regional RSL variability. In this study, we assess the importance of different contributions to RSL variability along the coast of Norway in the period 1960–2010. By using hydrographic station data at the coast, sea level pressure, and observed vertical land uplift, we compute RSL changes due to thermal expansion, haline contraction, the inverted barometer effect, and land uplift caused by glacial isostatic adjustment. The combination of these contributions is compared to RSL variability observed with tide gauges. For all but the two southernmost stations, the reconstructed RSL explains 70–85% of the observed variability of the monthly sampled time series. The inverted barometer effect is responsible for more than half of the explained variability, while thermosteric height represents the largest contribution to the linear trend. Due to land uplift, the local RSL rise is weaker and partly negative along the Norwegian coast. The residual (observed minus reconstructed) shows a positive trend ranging from 1.3 mm yr1 to 2.3 mm yr1. It is speculated that the reason for this is an increase of mass in the ocean due to melting of land-based ice and, to a lesser degree, the combined thermohaline expansion in the deep Nordic seas. Citation: Richter, K., J. E. Ø. Nilsen, and H. Drange (2012), Contributions to sea level variability along the Norwegian coast for 1960–2010, J. Geophys. Res., 117, C05038, doi:10.1029/2011JC007826.

1. Introduction [3] A main element of the rising sea level is the enhanced uptake of heat by the ocean and the subsequent thermal [2] Data from tide gauges indicate that the global sea level expansion of the water masses. Other major contributions has been rising by about 25 cm since the start of the obser- are the melting of land-based glaciers and the large ice sheets vational record in 1860 [Church and White, 2011; on Greenland and Antarctica. According to the Fourth Woodworth et al., 2011a]. Sea level rise since 1992, inferred Assessment Report (AR4) of the Intergovernmental Panel on from satellite-borne altimeters, shows an increase of slightly Climate Change (IPCC), thermal expansion contributed with more than 3 mm yr 1 [Cazenave and Llovel, 2010]. The two about 40% while glaciers and ice sheets accounted for the independent time series are in general agreement during the remaining 40% and 20%, respectively, for the period 1961– overlapping period [Woodworth et al., 2011a]. It is therefore 2003 [Meehl et al., 2007]. It is, however, only in recent years a well-established fact that the global sea level is rising. with improved observation platforms that the major con- Whether the rise of the observed sea level is accelerating, tributing factors to the global sea level form a near closed which is a key question related to projections of future sea budget [Cazenave et al., 2008; Leuliette and Miller, 2009]. level rise and its consequences for coastal regions, depends [4] Recent analyses show that the relative importance of heavily on the time span of the analysis. Based on a recon- the individual contributions to the global sea level is subject structed sea level time series, both Jevrejeva et al. [2006] to large temporal variations. Synthesis studies by Cazenave and Woodworth et al. [2011b] indicate an accelerated sea et al. [2008] and Cazenave and Llovel [2010] suggest that level rise starting at the end of the 19th century. the importance of melting ice sheets to the total sea level rise increased to 40% during the period 2003–2007/2008, whereas the contribution from thermal expansion decreased 1Geophysical Institute, University of Bergen, Bergen, Norway. 2Bjerknes Center for Climate Research, Bergen, Norway. to about 15%. The increasing contribution from melting ice 3Nansen Environmental and Remote Sensing Center, Bergen, Norway. sheets is in general agreement with the reported acceleration 4Uni Research AS, Bergen, Norway. of the melting of land-based ice [Rignot et al., 2011]. [5] Sea level is not rising at the same rate globally but Corresponding author: K. Richter, Geophysical Institute, University of exhibits significant spatial variations [Church et al., 2004; Bergen, Allégaten 70, N-5007 Bergen, Norway. ([email protected]) Cazenave et al., 2008] in addition to the large interannual Copyright 2012 by the American Geophysical Union. and decadal fluctuations superimposed on the long-term 0148-0227/12/2011JC007826

C05038 1of12 C05038 RICHTER ET AL.: SEA LEVEL ALONG THE NORWEGIAN COAST C05038 increasing trend [Jevrejeva et al., 2006]. Several studies respect to land do not necessarily reflect changes in sea level have examined local changes in the observed, and to some in the open ocean. In this study, we investigate changes in extent in the modeled, sea level. Examples of regional sea the relative sea level (RSL), that is changes in the sea level level analysis cover the Arctic Ocean [Pavlov, 2001; with respect to land. For sea level estimates representing the Proshutinsky et al., 2001, 2004, 2007], the North Atlantic open ocean, we use the term sea surface height (SSH). Ocean [Llovel et al., 2011], the coast of Scotland [Rennie [10] Among other, less directly observable contributions and Hansom, 2011], the German Bight [Wahl et al., 2011], to local RSL change are redistribution of ocean mass onto the Mediterranean Sea [García-García et al., 2010; continental shelves due to deep ocean expansion, and the Meyssignac et al., 2011], and the coast of the U.S.A. [Yin expectedly important mass input from glaciers and ice caps. et al., 2009; Weiss et al., 2011]. Recent regional sea level These contributions will be roughly estimated and briefly studies deduced from global analysis are given by, e.g., Riva discussed in light of the observed RSL variability and trends. et al. [2010], Yin et al. [2010], Pardaens et al. [2011], [11] The main objective of the present study is to identify Marcos et al. [2011] and Slangen et al. [2011]. and quantify the contributions to the observed RSL vari- [6] The objective of this study is to quantify, to the extent ability along the Norwegian coast, and to assess to what possible, the major contributions to the observed sea level degree RSL can be estimated from the contributions taken changes, in terms of variability and trends, along the eastern into account. rim of the Nordic seas during the past 50 years. While the [12] In section 2 we describe the data and methods used to steric height anomaly is globally mostly accounted for by compute and analyze the contributions to RSL variability. thermal expansion, haline contraction is almost equally Results are presented and observed RSL is compared to its important in the subpolar North Atlantic. Here, cooling and reconstruction in section 3. The outcome and missing con- simultaneous freshening lead to density compensated linear tributions are discussed in section 4 and the study is con- trends in steric height during the second half of the 20th cluded in section 5. century [Antonov et al., 2002; Levitus et al., 2005; Steele and Ermold, 2007; Siegismund et al., 2007]. It is however 2. Data and Methods not clear to what extent changes in sea surface height in the interior of the Nordic seas are relevant for changes in sea [13] Various factors contribute to variations in RSL. level along the Norwegian coast. The coastal water is, for In this study, we consider changes induced by atmospheric instance, influenced by both relatively warm Atlantic Water and thermohaline variability, as well as vertical land uplift. and fresh water from land runoff, whereas the interior of the The combination of these contributions yields the recon- basin mainly consists of cold and relatively fresh water of structed RSL, polar origin. ¼ h þ h þ h þ ; ð Þ RSLrc p T S GIA 1 [7] The dominant mode of atmospheric variability over the Nordic seas is characterized by a meridional pressure which will be compared to the observed RSL. In the gradient and southwesterly winds, commonly represented by expression above, h is the SSH variability due to surface the North Atlantic Oscillation (NAO) index [Hurrell, 1995]. p pressure fluctuations (the IBE effect), hT and hS are the The effect of atmospheric forcing on sea level along the thermosteric and halosteric contributions, respectively, and Norwegian coast is dual. Changes in surface pressure affect GIA is a linear trend representing vertical land uplift due to sea surface height through the inverted barometer effect glacial isostatic adjustment. Accordingly, (IBE) while the prevailing southwesterly winds push the ¼ þ h ; ð Þ water onshore. Storm surges combine both effects and may RSL RSLrc res 2 lead to exceptionally high sea levels along the northwestern European coast, impacting existing infrastructures. where hres is the sea level residual that is not explained by [8] Wakelin et al. [2003] found good correlation between our reconstruction. sea level observed along the western European coast and the [14] In the following, we will describe the data we use to NAO index. In particular, wind effects appeared to dominate compute the single contributions, and how we quantify their over IBE in the southeastern North Sea while the wind relative importance on interannual time scales and with contribution away from the shelf was negligible. Jevrejeva respect to trends. et al. [2005] demonstrated that the link between NAO and European winter sea level persisted only for selective time 2.1. Tide Gauge Data intervals, probably due to a meridional shift in westerlies [15] Observed RSL is obtained from the historical tide that is not properly represented by a static NAO index. This gauge data set compiled by the Permanent Service for Mean finding strongly indicates cautious use of the NAO index as Sea Level (PSMSL) [Woodworth and Player, 2003] for the a time invariant proxy for winter variations in the sea level in period 1960–2010. We use exclusively the Revised Local the region. Reference data, presenting the sea level measured relative to [9] Fennoscandinavia experiences relatively large, but a coastal benchmark at each station. Our criteria for selecting spatially nonuniform, glacial isostatic uplift [Ekman, 1996; tide gauge time series are multidecadal records (>30 years) Milne et al., 2001]. Recent estimates of the uplift rates are without large gaps (>1.5 years) covering the past 50 years given by Vestøl [2006], varying from more than 40 cm per (Figure 1 and Table 1). Longer records are available at some century in the inner Oslo fjord in southeast Norway to about stations but for consistency and due to the limited avail- 10 cm per century along the outermost part of southern and ability of auxiliary data (see below), we confine our analysis western Norway. Consequently, and particularly on multi- to the period after 1960. decadal and longer time scales, changes in sea level with

2of12 C05038 RICHTER ET AL.: SEA LEVEL ALONG THE NORWEGIAN COAST C05038

and global positioning system data to derive land uplift rates for Fennoscandinavia. [17] Land uplift estimates as predicted by the ICE-5G models are supplied by the Permanent Service for Mean Sea Level [Peltier, 2004] for two different Earth models (VM2 and VM4). There are large discrepancies between the mod- eled estimates (not shown). Both are lower than the esti- mates from Vestøl [2006]. Peltiers model is global, so small- scale anomalies in Earths structure are not properly modeled. [18] Therefore, we use the data set provided by Vestøl [2006]. It is partly based on tide gauges (from longer peri- ods than discussed here) but we consider it the best available estimate of vertical land uplift rates to date. Rates of vertical uplift at the positions of the tide gauge stations are presented in Table 1. [19] A problem here is that data coverage close to tide gauge stations is in general sparse. The reliability of the uplift rates varies geographically depending on the spatial density of tide gauges with long-term observations of sea level. Thus, uncertainties in the derived rates are not geo- graphically uniform [Vestøl, 2006, Figure 7]. In this study, however, we use the average uncertainty of 0.5 mm yr1. 2.3. Thermohaline Contributions [20] Hydrographic station data along the Norwegian coast were provided by the Institute of Marine Research (IMR), Bergen, Norway. There are eight permanent stations along the Norwegian coast (Figure 1) that have been maintained Figure 1. Positions of the tide gauges used in the analysis for several decades and provide vertical profiles of temper- (black) and locations of the hydrographic stations (blue). ature and salinity. The frequency of measurements varies with time. Except for Skrova at 68N with almost weekly sampling, each station has been sampled approximately 2.2. Vertical Land Movement twice a month, but with gaps in the data from one to several [16] The most important process contributing to vertical months. Data cover the period 1960–2010, with the excep- land movement in the northeastern North Atlantic is GIA (as tion of Ingøy (1968–2010) and Bud (1971–2010). opposed to earthquakes, increased groundwater extraction [21] Steric height along the coast is computed following and deposits from river discharges). There are two approa- McClimans et al. [1999]: ches to determine vertical land uplifts related to GIA. The Z most common approach is by means of geodynamic mod- h ¼ ðr rÞ=r dz; ð3Þ eling. The second approach is to use observations. Vestøl st 0 0 [2006] combined leveling, historical tide gauge recordings

Table 1. Tide Gauges With Monthly Observations of RSL Obtained From the PSMSLa Tide Gauge Longitude Latitude Period GIA (mm yr1) IMR Station Honningsvåg 25.6 70.6 1970– 2.1 Ingøy Hammerfest 23.4 70.4 1957– 2.5 Ingøy Tromsø 18.6 69.4 1952– 2.7 Skrova Harstad 16.3 68.5 1952– 2.7 Skrova Narvik 17.2 68.3 1928–1940, 1947– 4.4 Skrova Kabelvåg 14.3 68.1 1948– 2.6 Skrova Bodø 14.2 67.2 1949– 3.6 Skrova Rørvik 11.2 64.5 1972– 4.2 Bud Heimsjø 9.1 63.3 1928– 3.1 Bud Kristiansund 7.5 63.1 1952– 2.6 Bud Ålesund 6.1 62.3 1945–1946, 1951– 1.9 Sognesjøen Måløy 5.1 61.6 1943– 1.9 Sognesjøen Bergen 5.2 60.2 1883–1889, 1928– 1.7 Sognesjøen Stavanger 5.4 58.6 1919–1939, 1946– 1.2 Indre Utsira Tregde 7.3 58.0 1927– 1.3 Lista Oslo 10.4 59.5 1885–1890, 1914– 4.9 Lista aThe position of each tide gauge station and the availability of data is given. GIA is provided in the fifth column based on Vestøl [2006]. The last column shows the hydrographic station used to compute the thermohaline contributions for the selected tide gauges (see Figure 1). The stations in bold are discussed in section 3.2.

3of12 C05038 RICHTER ET AL.: SEA LEVEL ALONG THE NORWEGIAN COAST C05038 where the integration is from the bottom of the profile to the [25] For two time series x and y, the correlation coefficient surface. The reference density r0 is the mean of the vertical is computed using average of all profiles at the given station. Absolute steric P r xy covðx; yÞ height is very sensitive to the choice of 0, while anomalies r ¼ pPffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiP ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : ð6Þ and linear trends are rather unaffected. According to Gill and x2 y2 varðxÞvarðyÞ Niiler [1973], hst can be divided into a thermal (hT) and a haline (hS) component assuming the deviations from a ref- The variance explained by RSLrc in equation (2) is then erence temperature and salinity are small. Accordingly, varðh Þ Z Z R2 ¼ 1 res * * * * varðRSLÞ h ¼ aðT ; S ÞðT T0Þ dz; h ¼ bðT ; S ÞðS S0Þdz; T S varð RSL h h h Þ ¼ 1 p T S ; ð7Þ ð4Þ varð RSLÞ where a and b are the thermal expansion and haline con- while the variances explained by the contributions in traction coefficients of sea water, respectively [McDougall, equation (1) are * * 1987], evaluated at T =(T + T0)/2 and S =(S + S0)/2 fol- varðRSL h Þ lowing Siegismund et al. [2007]. Here, T0 and S0 are refer- R2 ¼ 1 i : ð8Þ ence temperature and salinity, respectively. i varðRSLÞ [22] The locations of the hydrographic stations are not h identical to the locations of tide gauges (Figure 1). There- However, as the i are not independent fore, RSL observations from tide gauges have been paired X 2 ≠ 2; ð Þ with the steric height, computed from (3), based on their R Ri 9 location and the highest correlation coefficients between i steric height and RSL observed with tide gauges (not dis- and comparison of equation (8) with equation (7) shows that played). Table 1 shows which hydrographic station has been ! assigned to which tide gauge station. The hydrographic X X covðh ; h Þ stations at Ytre Utsira and Indre Utsira, and at Skrova and R2 ¼ R2 i j : ð10Þ i varðRSLÞ Eggum are situated close to each other (Figure 1). Accord- i j≠i ingly, from each pair only one station, namely Indre Utsira and Skrova respectively, has been used for the analysis. Thus, covariances between the predictors are important as they contribute to the explained variance. 2.4. Inverted Barometer Effect [26] In addition to covariances, the observed RSL and its [23] The SSH variability is strongly influenced by changes contributions were analyzed with respect to linear trends. To in atmospheric pressure through the inverted barometer remove contributions from short-term variability and the effect (IBE). Generally, a 1 mbar increase in surface pressure seasonal cycle, trends were computed from data low passed produces a 1 cm depression of sea level. We use monthly with a 1 year running mean. Prior to filtering the data, gaps atmospheric surface pressure from the National Centers for were filled using the seasonal cycle. After filtering, the gaps Environmental Prediction-National Center for Atmospheric were reinserted and trends have been obtained by linear Research (NCEP-NCAR) reanalysis [Kalnay et al., 1996] regression using least squares. The respective uncertainties at 2.5 spatial resolution to apply the inverted barometric are given by the 95% confidence intervals of the regression correction coefficients. [27] To assess the trend of the residual sea level, that is the h ¼D =ðr Þ: ð Þ p p 0g 5 sea level that is not explained by vertical land uplift and changes in surface pressure and hydrography, the following Here Dp is the pressure fluctuations leading to IBE, r0 is the linear model is assumed reference density of sea water taken as 1025 kg m3, and g is h ¼ b þ b þ b ðh þ h þ h Þþ ð Þ the acceleration due to gravity. The pressure fluctuations are 0 1t 2 p T S 11 defined as the deviations from the mean over the period 1960–2010. The pressure is taken from the ocean grid point b ¼ b* : ð Þ closest to the location of the tide gauge station. 1 1 GIA 12

2.5. Variability and Trends Here, h is the observed RSL and bi are the coefficients h [24] The focus of this study is to assess the contribution to obtained by regressing on time t and the sum of the com- the above mentioned components to the observed RSL var- ponents, and minimizing the error term using the least iability and change. The analysis is based on monthly time squares method. As GIA represents a pure linear trend with a series. Covariability is explored by means of correlation and relatively large uncertainty, it is treated separately from the b covariance analysis. If not stated otherwise, the linear trend other components. In the above expressions, 0 is the b has been removed from all time series prior to computing intercept of the model, and 2 is a measure for how much of correlation coefficients and covariances. Therefore, linear the variability of h is explained by the sum of steric and long-term trends are not included in the covariance analysis barometric contributions and is close to 1 if those contribu- b although they may contribute significantly to the variance on tions account for most of the variability. 1 represents a long time scales. linear trend in RSL and includes vertical land uplift as well

4of12 C05038 RICHTER ET AL.: SEA LEVEL ALONG THE NORWEGIAN COAST C05038

Figure 2. The leading mode of variability of observed RSL along the Norwegian coast (black), and the leading mode of Figure 3. Trends of relative sea level (RSL, black) and sea SSH, i.e., RSL corrected for vertical land uplift (gray). surface height (SSH, red) from the period 1960–2010 in mm Explained variances are 65% for RSL and 83% for SSH yr 1 for tide gauges presented in Table 1. data.

* * as the residual (unexplained) trend, b1. b1 represents the close to zero elsewhere. At Oslo, RSL is sinking at a rate of * 1 trend in SSH and is obtained by b1 = b1 + GIA. The 2.5 mm yr . After correcting for vertical land uplift, the respective uncertainties are computed from the square root trends are significantly larger and positive at all stations of the summed square errors. (>1.7 mm yr1, with a mean value of 2.6 mm yr1). Thus, [28] To explore the common variability of all observa- vertical land uplift substantially weakens RSL rise along tions, we use empirical orthogonal function (EOF) analysis the Norwegian coast. and present the corresponding principal components (PC). [31] The leading EOF mode of SSH (Figure 2) features a For vector fields such as surface wind, the components u and positive trend starting around 1985, indicating that rates v are treated as two fields and the EOF analysis is performed of SSH rise were comparable to or exceed rates of vertical on the joint matrix (u, v). land uplift sometime during the 1980s. The linear trend in the SSH-EOF for the whole period is 2.9 0.3 mm yr1, 3. Results or 3.3 mm yr 1 larger than the trend in RSL. 3.1. Relative Sea Level and Sea Surface Height [29] The RSL observations used in this study are spread along the Norwegian coast, thus spanning more than 10 latitudinal degrees. To extract the variability common to all stations, we perform an EOF analysis on the observations for the period 1960–2010. Gaps in the data are zero padded and all time series are low passed with a 1 year running mean filter prior to performing the analysis, in order to exclude contributions from the seasonal cycle. The leading mode of variability (Figure 2) explains 65% of the observed vari- ance. The corresponding spatial pattern (not displayed) shows that this mode is most important in Oslo and north of Ålesund and less significant (but still positive) from Tregde to Måløy. The displayed RSL based on the first EOF is dominated by large interannual variability, particularly in the 1980s and early 1990s. A negative trend is seen in the first part of the record but it levels off sometime during the 1980s. The overall trend is 0.4 0.3 mm yr1 and there- fore just significantly different from zero. [30] To assess the effect of vertical land uplift given in Table 1, we compute linear trends from the observations of RSL for the period 1960–2010 (where available) and com- pare the results with trends corrected for land uplift, i.e., Figure 4. Contributions from land uplift, IBE, and ther- trends in SSH, in Figure 3. The RSL trends are spatially not mosteric and halosteric heights to RSL variability in Bergen. uniform along the coast of Norway. They are positive from For better visualization, the monthly time series have been Tregde to Ålesund and north of Harstad, and negative or low passed by applying a 1 year running mean.

5of12 C05038 RICHTER ET AL.: SEA LEVEL ALONG THE NORWEGIAN COAST C05038

Figure 5. (top) Explained variances according to equations (7) and (8) for (a) high-pass-filtered and (b) low-pass-filtered data. Cutoff frequency is 1 year. Colored bars refer to the contributions and dots to their combination. (bottom) Covariances for (c) high-passed and (d) low-passed data. The covariances are normalized with var(RSL) and multiplied with a factor of 2 in accordance with equation (10). Thus, the difference between the sums of the bars in Figures 5a and 5c and Figures 5b and 5d equal the dots in Figures 5a and 5b, respectively. Covariances framed with a thick black line are statistically significant at the 99% confidence level.

3.2. Contributions to Relative Sea Level of 1 year. Note that the trends are not included in the [32] In this study, RSL at the tide gauge locations pre- covariance analysis. Depending on the location, the recon- sented in Table 1 is reconstructed by computing and com- struction explains 28% to 86% of the observed intra-annual bining observed land uplift with SSH fluctuations induced variability, and 38% to 75% of the observed interannual by the IBE and thermosteric and halosteric height following variability. The hydrographic station at Lista (Figure 1) may equation (1). Figure 4 shows these contributions to the not be representative for the hydrography in the Oslofjord interannual (1 year running mean applied) RSL variability since it is remotely located. If Oslo is excluded, the mini- for Bergen. The IBE explains most of the variance (47%), mum variance explained by our reconstruction is 45% on while the thermosteric and halosteric components account intra-annual and 48% on interannual time scales at Tregde. for about 12% and 18%, respectively. Combined, they [34] The sums of the explained variances of the contribu- explain 62% of the observed interannual RSL variability tions (color bars in Figures 5a and 5b) exceed the explained in Bergen. Both the IBE and halosteric height are dominated variance of the reconstructed RSL (dots) indicating cov- by strong interannual variability with no obvious trends. ariability between the contributions (Figures 5c and 5d). In contrast, thermosteric height features low-frequency, Indeed, there are significant and positive covariances decadal variation and a positive trend in the second half of between IBE and thermosteric and halosteric height on intra- the period. annual time scales. Covariances between IBE and halosteric [33] The explained variances of the single contributions as height are strongest on intra-annual time scales, but are also well as of their combination at all stations are summarized present on interannual time scales. In addition, there is a and presented in Figure 5. The analysis was performed for tendency toward density compensation on longer time short-term and long-term variability, with a cutoff frequency scales. The latter is, however, only significant at the two southernmost stations.

6of12 C05038 RICHTER ET AL.: SEA LEVEL ALONG THE NORWEGIAN COAST C05038

Figure 6. (a) Mean spectral density in cm2 cpy1 (cycles per year) of observations (black), reconstruc- tion (gray), and contributions from the IBE (green) and thermosteric (red) and halosteric (blue) heights. The data have been filtered with a 256 point Hanning window prior to computing the spectra. (b) Mean seasonal cycle.

[35] Except for the tide gauges in Oslo and Tregde, the from halosteric height to the seasonal cycle is considerably contribution from the IBE is largest on both time scales. On smaller with a minimum centered around July and a maximum intra-annual time scales, halosteric height equals or exceeds in November. The sum of all contributions has a mean sea- the contribution of thermosteric height hinting toward the sonal cycle with an amplitude of about 9 cm and a minimum importance of the relatively fresh water carried along the (maximum) in May (November) while the amplitude in the Norwegian coast by the Norwegian Coastal Current. On observations is 12 cm with minimum in April/May and max- longer time scales, the relative contribution from halosteric imum in November/December. height decreases. Apart from the southernmost stations Oslo [39] Figure 7 presents the trends of the IBE and steric and Tregde, both thermosteric and halosteric height con- contributions. Comparison with rates of vertical land uplift tribute positively to the explained variance. Hence, there is (Table 1) shows that the uplift dominates the linear trend of in general no net density compensation present in the coastal the contributions. Vertical land uplift aside, the largest trend water column. The explained variance of the reconstructed contribution is from the thermosteric component, ranging RSL decreases at interannual time scales (Figure 5a versus from 0.5 mm yr1 in Stavanger (Indre Utsira) to 1.0 mm Figure 5b), indicating that other processes are more impor- yr1 in Kristiansund, where it has been computed over a tant on longer time scales. shorter period, and Tregde. This result indicates a net [36] The explained variances for the unfiltered time series warming of the water column along the entire Norwegian (not shown) are very similar to the explained variances for shelf. The halosteric contribution compensates with weak the high-pass-filtered time series in Figure 5a with maximum negative or close to zero long-term trends that correspond to in Harstad (85%) and minimum in Oslo (30%). a net salinification of the water column. The exception is [37] Due to the sparseness of the hydrographic stations as well as the rather coarse resolution of the gridded surface pressure field, only a few tide gauge stations are unique in terms of the chosen contributing factors. Based on the results in Figure 5 and the length and completeness of the available data record, Tregde, Stavanger, Bergen, Kristiansund, Tromsø and Hammerfest are selected for the following analysis. [38] The temporal variability of observed RSL and its contributions is assessed by computing the respective spec- tra at each of the selected stations. The mean spectra are presented in Figure 6a. The IBE dominates on subseasonal time scales and is the least important contribution on decadal time scales. Except for a seasonal peak, its spectrum is white. The steric height spectra show increasing power with increasing periods. Common to all contributions is a distinct peak at the annual period. IBE and thermosteric height have on average the same amount of power in the seasonal cycle. However, the phase of the associated seasonal cycle varies (Figure 6b). While the IBE peaks in December, thermosteric Figure 7. Observed trends of IBE (black) and thermosteric height does so already in September/October. The contribution and halosteric heights (red and blue).

7of12 C05038 RICHTER ET AL.: SEA LEVEL ALONG THE NORWEGIAN COAST C05038

Figure 8. Observed (red) and reconstructed (black) RSL at selected stations. To improve the visualiza- tion and to emphasize interannual variability, the data has been low passed by using a 1 year running mean. Correlation coefficients between observations and reconstructions are presented.

Tromsø (Skrova) where freshwater appears to accumulate captured by the reconstruction. There are however occasional leading to a positive trend comparable to that of IBE. Trends deviations from the observations, e.g., in the late 80s from due to long-term changes in surface pressure are insignifi- Bergen to Tromsø. As these excursions are present at several cant except at the station in Tromsø where the IBE con- stations, it is unlikely that they are the result of erroneous tributes positively to the trend. observations but represent actual strong anomalies caused by other processes than those accounted for in this analysis. 3.3. Reconstruction and Residuals [41] The residual trend, i.e., the trend that is not accounted [40] Figure 8 presents RSL observations at the selected for by our RSL reconstruction, and its uncertainty for the six stations together with their unique reconstructions. As there is selected stations are presented in Figure 9. The trend is no data available at the hydrographic stations in Ingøy and positive for all stations and ranges from 1.3 mm yr1 at Bud prior to 1970 we cannot reconstruct RSL further back in Tregde to 2.3 mm yr1 at Hammerfest (Table 2). These time for Hammerfest and Kristiansund. Observed and recon- trends are comparable to or larger than the combined trend structed RSL agree well. The interannual variability is mostly owing to pressure and steric changes (Figure 7). Evidently, additional processes contribute significantly to the observed long-term trends in RSL along the Norwegian coast.

4. Discussion

[42] We have estimated RSL variability along the Norwegian coast at various locations by accounting for vertical land uplift, atmospheric loading and steric con- tributions. The combination of these effects was compared to historical observations from tide gauges and we found that, depending on the location, our estimate explains 30–85% of

Table 2. Output of Model as Described by Equations (11) and (12)

1 * 1 Station b0(mm) b1 (mm yr ) b2 b1 (mm yr ) Hammerfest 6934 0.2 1.1 2.3 0.7 Tromsø 6958 0.8 1.0 1.9 0.6 Kristiansund 7006 0.7 1.1 1.9 0.7 Bergen 6949 0.2 0.9 1.9 0.6 Figure 9. Residual trends, b*, computed from Stavanger 6925 0.6 1.0 1.8 0.6 1 Tregde 7010 0 0.7 1.3 0.6 equations (11) and (12) for the six selected stations.

8of12 C05038 RICHTER ET AL.: SEA LEVEL ALONG THE NORWEGIAN COAST C05038

of the Atlantic Meridional Overturning circulation (AMOC) result in regional dynamic sea level changes [Levermann et al., 2005; Yin et al., 2010]. This effect may alter the influx of Atlantic Water to the Nordic seas in a warmer world, but it is likely small for the time period considered here. In addition, there is no observation-based evidence for long-term changes in AMOC, also indicating that the men- tioned effect—for the present day climate—is small. [47] Other forcing mechanisms for variability and trend that are worthwhile discussing, are (i) the dominant atmo- spheric pattern of variability in the Nordic seas favors southwesterly winds that, through Ekman transport, push water toward the coast; (ii) warming and/or freshening of the deep ocean within the Nordic seas result in ocean mass redistribution, moving water from the interior onto the shallow shelf areas; (iii) melting of land based ice redis- tributes water from land to the oceans while the opposite is Figure 10. Principal components of the leading modes of true for the retainment of water, e.g., through storage. In the NCEP wind over the northern North Atlantic (black) and following sections these effects will be examined further. residual sea level (red). For clarity, the principal components 4.1. Effect of Wind have been low passed with a 1 year running mean. Explained variances are indicated in the legend. The inset shows the [48] In order to assess the impact of wind on observed normalized spatial wind pattern of the leading mode. Color RSL, we perform an EOF analysis of the wind over the coding is a measure for the strength of the wind amplitudes, northern North Atlantic and Nordic seas, and compare it ranging from 0 to 1 in intervals of 0.1. with the leading mode of the sea level residual (Figure 10). To also address the unexplained sea level variance on intra- annual time scales, monthly data are used. Surface wind is the observed variability. On subdecadal time scales the bulk obtained from the NCEP-NCAR reanalysis [Kalnay et al., of the observed variability is explained by the IBE. In par- 1996]. The following analysis has been duplicated using ticular, IBE and halosteric height covary significantly, thus surface wind stress instead of wind, with essentially identical amplifying each other. Anomalously low sea surface pres- results (not shown). sure (positive IBE anomaly) is often related to more storms [49] The leading mode of the wind field shows a pattern traveling into the area. Storms cause southwesterly wind with strong westerlies over the Irminger Sea and winds anomalies that result in eastward propagation and subse- parallel to a major portion of the northern European shelf. quent downwelling of the fresh coastal water. These The corresponding principal component features seasonal mechanisms are important on seasonal (Figures 5c and 6b) (not shown), interannual as well as decadal variability. The as well as interannual time scales (Figure 5d). latter includes a positive trend starting in the late 1960s [43] Owing to strong land uplift, the trend in RSL is persisting until the late 1980s. Since then, long-term fluc- reduced substantially along the entire Norwegian coast. tuations are moderate although superimposed by large However, rates of sea level rise appear to be large enough to interannual fluctuations. The correlation coefficients compensate for vertical land uplift, resulting in positive between PCs of wind and residual sea level are r = 0.42 on observed RSL trends along large portions of the coast. A monthly time scales and r = 0.19 for data low passed with a similar result has been reported by Rennie and Hansom 1 year running mean. The spectrum of the leading wind [2011] for the coast of Scotland where the land uplift is mode (Figure 11) has a peak at the annual period, suggesting comparable to the uplift in southern and western Norway. that the part of the seasonal cycle not explained by our [44] The effect of thermal expansion is evident at all reconstruction (Figure 6) is caused by wind forcing. Indeed, stations with trends of up to 1.0 mm yr1, thus dominating the seasonal cycle of the leading wind mode has a strong the trend budget compared to halosteric and surface pressure minimum in May (3ms 1) and a maximum in December induced long-term variability. It is worth noting that the and January (2 m s 1) (not shown). importance of the IBE trend increases toward higher lati- [50] In addition to the seasonal cycle, the spectra of lead- tudes. This is consistent with the reported decrease of sur- ing modes of wind and sea level residual have increased face pressure over the Arctic Ocean [Walsh et al., 1996]. power toward longer time scales (Figure 11), with the [45] The positive trend in the residual sea level indicates spectral density of the sea level residual exceeding those of that the observed trend is substantially underestimated by the wind on the longest time scale. The positive trend in our reconstruction. For the period 1960–2000, Marcos and 1970–1990 is common to both time series, indicating that the Tsimplis [2007] found residual trends of 1.1 mm yr1 in rise in RSL during this period is partly wind driven. How- the NE Atlantic and 1.3 mm yr1 in the North Sea. Those ever, the observed RSL continues to rise although there is no trends are comparable to the residual trends we found in apparent long-term trend in the wind forcing (Figure 10). The Southern Norway. There are several factors not included into overall trend in wind is close to zero (0.02 0.01 m s 1 our analysis that may contribute to a rise in observed RSL. yr 1). This hints to other processes being responsible for the [46] Changes in the circulation of the ocean contribute to observed rise, in particular during the last decade. sea level variability. Numerical models indicate that changes

9of12 C05038 RICHTER ET AL.: SEA LEVEL ALONG THE NORWEGIAN COAST C05038

[53] As a result of increased SSH in the interior Nordic seas, ocean mass is transferred toward the shallow shelf, contributing to the observed sea level rise along the coast. The time scale of the adjustment of the sea surface is in the order of a few days (surface gravity waves). However, changes in the deep ocean alter the density structure of the water column (lifting isotherms etc). This leads to internal adjustment processes that take place on time scales of years to decades. 4.3. Mass-Exchange-Related Trends on the Norwegian Coast [54] The world’s ice sheets, glaciers and ice caps have been losing mass at an increasing rate over the last decades. As opposed to the other effects discussed in this paper, which redistribute mass inside the oceans, this adds mass to the Figure 11. Spectral density in 1 cpy (cycles per year) of oceans. As a direct result, the global sea level rises according first principal component of wind (black) and sea level resid- to the mass received (eustatic sea level rise). However, the ual (red). The data have been standardized prior to comput- change in the earths gravitational field due of the loss of mass ing the spectra. Spectra were computed using a 256 point from the land sources, causes a redistribution of the sea level, fast Fourier transform with a Hanning window and 50% in addition to the causes already discussed. overlap. [55] In total, the mass loss from land ice has contributed to an eustatic sea level rise of 1.09 0.26 mm yr1 in the period 1972–2008 [Church et al., 2011]. This can be 4.2. Redistribution of Ocean Mass attributed to three main contributors: Greenland Ice Sheet 1 [51] Steric sea level changes in the interior Nordic seas (GIS) with 0.12 0.17 mm yr ; Antarctic Ice Sheet (AIS) 1 may affect SSH changes along the Norwegian coast through with 0.30 0.20 mm yr ; and glaciers and ice caps (GIC) 1 redistribution of ocean mass. In particular, changes in the with 0.67 0.03 mm yr . thermohaline structure of the deep ocean generate horizontal [56] When mass is lost from a region, such as an ice sheet pressure gradients at the surface with respect to the shallow or glacier, the (horizontal) gravitational pull from that region shelf areas and ocean mass is transferred to balance the is weakened and water levels around the source sinks. This gradients [e.g., Yin et al., 2010]. Warming of the deep abyss effect is not negligible. As explained by Tamisiea et al. has been reported by Østerhus and Gammelsrød [1999] and [2003], the gravitational change results in a redistribution attributed to the variability in the exchanges between the of global sea level equal to the exchanged mass multiplied three deep basins of the Nordic seas. In addition, the by factors ranging from below zero to above one. These warming signal observed in the upper ocean [Holliday et al., patterns are called fingerprints of the source in question, and 2008] may gradually penetrate into the deep ocean leading reach around the globe with the strongest (diminishing) to even stronger warming there. effect close to the source and above eustatic sea level rise in [52] Figure 12 shows changes in steric height computed in far away regions. There is also a smaller and slower but also the 500–2000 m depth range at Ocean Weather Station Mike (OWSM, 66N, 2E) for the period 1960–2006. At these depths, the water masses are very cold and density follows mostly salinity. This translates to the steric height being governed by its halosteric component. However, thermos- teric height tends to follow the steric height as well, i.e., as the water gets fresher it gets also warmer. Prior to the 1980s the steric height in the deep ocean features a weak negative trend but is mostly dominated by variability on shorter time scales. Around 1990, steric height increases abruptly at OWSM by roughly 1 cm within only 2 years. The fast increase is mostly accounted for by the halosteric compo- nent, i.e., freshening of the deep water whereas the ther- mosteric component increases more moderately. Steric height continues to increase steadily though at a lower rate than observed in 1990. The jump is due to warmer and fresher intermediate waters reaching OWSM and it is rea- sonable to assume that the filling up of the whole Nordic seas has been a gradual process since around 1980 [Østerhus Figure 12. Steric height anomalies (black) computed and Gammelsrød, 1999]. The overall trend for the 1960– between 500 and 2000 m from hydrographic profiles at 2006 period is 0.4 0.1 mm yr1 while, for the period OWSM. Thermosteric (red) and halosteric (blue) compo- 1980–2006, the increase is 1.3 0.1 mm yr1. nents are shown as well. The long-term mean was removed, and a 1 year running mean was applied.

10 of 12 C05038 RICHTER ET AL.: SEA LEVEL ALONG THE NORWEGIAN COAST C05038 global reaching crustal effect included in this. Factors for the to variability comes from IBE. For the steric effects, there are Norwegian coast, related to the three mentioned ice sources, large regional differences along the coast. can be deduced from the fingerprint patterns provided by [65] Linear trends in RSL are positive in southwestern and Mitrovica et al. [2001]: fGIS ≈ 0.0; fAIS ≈ 1.0; and fGIC ranges northern Norway. The most prominent contributions to the from 0.7 to 0.5 from southern to northern Norway. Due to trend are the land uplift and a positive thermosteric contri- the large uncertainties in both the mass loss estimates above bution from a warming in the coastal waters (e.g., –1.7 and and our residual trends (Figure 9), we will not be concerned 0.9 mm yr1, respectively, for Bergen). with the spatial differences along the Norwegian coast and [66] The mentioned contributions explain less than half of use a representative value of fGIC ≈ 0.6. the observed trend. One candidate for uncertainty is the ver- [57] The regional result of these three mass balance con- tical uplift rates. Other contributions not taken into account, tributions can then be estimated as but discussed, are wind effects, melting of land based ice as well as ocean mass redistribution due to hydrographic chan- b* ¼ D þ D þ D ; ð Þ MB hGIS fGIS hAIS fAIS hGIC fGIC 13 ges in the deep ocean. While being important on seasonal and interannual time scales, the effect of wind on the long-term where Dh represent the three eustatic sea level trends given trend appears to be minor. Likewise, little information about above. the melting ice sheets and glaciers is known prior to the [58] Error estimates for the fingerprint patterns are not advent of satellites, and modern estimates still vary largely. provided in the literature, likely on the account of errors in We estimate that melting of land-based ice has resulted in a the melt rates being much larger. And indeed, for our case, rate of sea level rise of about 0.7 0.2 mm yr1 along the the large uncertainty in the AIS mass loss dominates the Norwegian coast, corresponding to about one third of the formal propagation of errors through (13), giving an error of unexplained trend. In addition, water masses in the deep 1 b* 1 0.2 mm yr for MB. With respect to any influence from Nordic seas appear to expand at a rate of 0.4 0.1 mm yr . error in the fingerprint factors it would have to be as large as It is however unclear, how this increase will affect RSL along 0.2, which is unlikely judging from the level of detail in the the coast as it will also induce changes in the ocean circula- fingerprint patterns presented in Mitrovica et al. [2001]. tion. Common to these additional effects is an increase since [59] The resulting mass loss contribution to sea level trend the 1980s, which is also seen in our unexplained residual on the Norwegian coast for 1972–2008 is then 0.7 0.2 mm series. While the large-scale winds do not contribute to a yr 1. This value is comparable with the independent positive trend through the last 2 or 3 decades, the accelerating (GRACE) gravity measurements of sea level change due to rates of loss of land-based ice may explain part of the continental ice melt during 2003–2009 of 0.6 0.2 mm yr 1 remaining residual. In addition, changes in hydrology may for the same region [Riva et al., 2010]. contribute to the residual trend, but are also related to large [60] In comparison with our residual trend of 1.9 uncertainties. 1 0.6 mm yr for the Bergen case, the loss of land-based ice [67] Continued observations, in combination with a explains about one third of the residual trend. Although detailed numerical model, are likely needed to significantly this mass balance contribution to the Norwegian coast is improve our understanding of variations and changes in for a different epoch than the 1960–2010 period used in regional and local sea level. the residual trend estimation herein (Figure 9), comparison is relevant. [68] Acknowledgments. Hydrographic stations were obtained from the Institute of Marine Research, Bergen. We are grateful to Svein Østerhus [61] In addition to mass exchange from land based ice, for providing the data from the Ocean Weather Station Mike and to Laurent effects such as retention of liquid water also comes into play, Bertino and David Stephenson for useful discussions. The comments of in the same manner, and the GRACE-measured total mass- two anonymous reviewers helped to improve the quality of the paper. exchange-related sea level trend for the Norwegian coast is This work has received financial support from Bergen Kommune, the 1 EU FP7 MONARCH-A project (grant 242446), and the Centre for Climate 0.8 0.4 mm yr during 2003–2009 [Riva et al., 2010]. Dynamics at the Bjerknes Centre. This is contribution A396 from the This is a more relevant number to compare our residual trend Bjerknes Centre for Climate Research. to, but valid only for a far shorter period. References 5. Summary and Conclusion Antonov, J. I., S. Levitus, and T. P. Boyer (2002), Steric sea level variations during 1957–1994: Importance of salinity, J. Geophys. Res., 107(C12), [62] Natural and anthropogenically induced changes in 8013, doi:10.1029/2001JC000964. RSL have strong implications for all coastal communities Cazenave, A., and W. Llovel (2010), Contemporary sea level rise, Annu. including those along the long coast of Norway. To project Rev. Mar. Sci., 2, 145–173. Cazenave, A., A. Lombard, and W. Llovel (2008), Present-day sea level future changes in RSL it is essential to identify and quantify rise: A synthesis, C. R. Geosci., 340, 761–770. the different contributions to RSL variability and change. Church, J. A., and N. J. White (2011), Sea-level rise from the late 19th to – [63] Subdecadal variability in RSL along the Norwegian the early 21st century, Surv. Geophys., 32, 585 602, doi:10.1007/ s10712-011-9119-1. coast has been examined by comparing tide gauge observa- Church, J. A., N. J. White, R. Coleman, K. Lambeck, and J. X. Mitrovica tions of RSL with hydrographic measurements (steric height), (2004), Estimates of the regional distribution of sea level rise over the atmospheric surface pressure (IBE) as well as land uplift rates 1950–2000 period, J. Clim., 17, 2609–2625. ’ (GIA) for the period 1960–2010. Church, J. A., et al. (2011), Revisiting the Earth s sea-level and energy bud- gets from 1961 to 2008, Geophys. Res. Lett., 38, L18601, doi:10.1029/ [64] It is found that these components account for 30–85% 2011GL048794. of the observed variability on monthly to interannual time Ekman, M. (1996), A consistent map of the postglacial uplift of Fennoscandia, scales, depending on location, and that the largest contribution Terra Nova, 8,158–165.

11 of 12 C05038 RICHTER ET AL.: SEA LEVEL ALONG THE NORWEGIAN COAST C05038

García-García, D., B. F. Chao, and J.-P. Boy (2010), Steric and mass- Peltier, W. R. (2004), Global glacial isostasy and the surface of the ice-age induced sea level variations in the Mediterranean Sea revisited, J. Geophys. Earth: The ICE-5G(VM2) model and GRACE, Annu. Rev. Earth Planet. Res., 115, C12016, doi:10.1029/2009JC005928. Sci., 32, 111–149. Gill, A. E., and P. P. Niiler (1973), The theory of the seasonal variability in Proshutinsky, A., V. Pavlov, and R. H. Bourke (2001), Sea level rise in the the ocean, Deep Sea Res. Oceanogr. Abstr., 20, 141–177. Arctic Ocean, Geophys. Res. Lett., 28, 2237–2240, doi:10.1029/ Holliday, N. P., et al. (2008), Reversal of the 1960s to 1990s freshening 2000GL012760. trend in the northeast North Atlantic and Nordic Seas, Geophys. Res. Proshutinsky, A., I. M. Ashik, E. N. Dvorkin, S. Häkkinen, R. A. Krishfield, Lett., 35, L03614, doi:10.1029/2007GL032675. and W. R. Peltier (2004), Secular sea level change in the Russian sector of Hurrell, J. W. (1995), Decadal trends in the North Atlantic Oscillation: the Arctic Ocean, J. Geophys. Res., 109, C03042, doi:10.1029/ Regional temperatures and precipitation, Science, 269(5224), 676–679. 2003JC002007. Jevrejeva, S., J. C. Moore, P. L. Woodworth, and A. Grinsted (2005), Influ- Proshutinsky, A., I. Ashik, S. Häkkinen, E. Hunke, R. Krishfield, M. Maltrud, ence of large-scale atmospheric circulation on European sea level: Results W. Maslowski, and J. Zhang (2007), Sea level variability in the Arctic based on the wavelet transform method, Tellus, Ser. A, 57, 183–193. Ocean from AOMIP models, J. Geophys. Res., 112, C04S08, Jevrejeva, S., A. Grinsted, J. C. Moore, and S. Holgate (2006), Nonlinear doi:10.1029/2006JC003916. trends and multiyear cycles in sea level records, J. Geophys. Res., 111, Rennie, A., and J. Hansom (2011), Sea level trend reversal: Land uplift C09012, doi:10.1029/2005JC003229. outpaced by sea level rise on Scotland’s coast, Geomorphology, 125, Kalnay, E., et al. (1996), The NCEP/NCAR 40-year reanalysis project, Bull. 193–202. Am. Meteorol. Soc., 77, 437–471. Rignot, E., I. Velicogna, M. van den Broeke, A. Monaghan, and J. Lenaerts Leuliette, E. W., and L. Miller (2009), Closing the sea level rise budget with (2011), Acceleration of the contribution of the Greenland and Antarctic altimetry, Argo, and GRACE, Geophys. Res. Lett., 36, L04608, ice sheets to sea level rise, Geophys. Res. Lett., 38, L05503, doi:10.1029/ doi:10.1029/2008GL036010. 2011GL046583. Levermann, A., A. Griesel, M. Hofmann, M. Montoya, and S. Rahmstorf Riva, R. E. M., J. L. Bamber, D. A. Lavallee, and B. Wouters (2010), Sea- (2005), Dynamic sea level changes following changes in the thermoha- level fingerprint of continental water and ice mass change from GRACE, line circulation, Climate Dynamics, 24, 347–354. Geophys. Res. Lett., 37, L19605, doi:10.1029/2010GL044770. Levitus, S., J. I. Antonov, T. P. Boyer, H. E. Garcia, and R. A. Locarnini Siegismund, F., J. Johannessen, H. Drange, K. A. Mork, and A. Korablev (2005), Linear trends of zonally averaged thermosteric, halosteric, and (2007), Steric height variability in the Nordic seas, J. Geophys. Res., total steric sea level for individual ocean basins and the world ocean, 112, C12010, doi:10.1029/2007JC004221. (1955–1959)–(1994–1998), Geophys. Res. Lett., 32, L16601, Slangen, A. B. A., C. A. Katsman, R. S. W. van de Wal, L. L. A. Vermeersen, doi:10.1029/2005GL023761. and R. E. M. Riva (2011), Towards regional projections of twenty-first Llovel, W., B. Meyssignac, and A. Cazenave (2011), Steric sea level varia- century sea-level change based on IPCC SRES scenarios, Clim. Dyn., 38, tions over 2004–2010 as a function of region and depth: Inference on the 1191–1209. mass component variability in the North Atlantic Ocean, Geophys. Res. Steele, M., and W. Ermold (2007), Steric sea level change in the Northern Lett., 38, L15608, doi:10.1029/2011GL047411. Seas, J. Clim., 20, 403–417. Marcos, M., and M. N. Tsimplis (2007), Forcing of coastal sea level rise Tamisiea, M. E., J. X. Mitrovica, J. L. Davis, and G. A. Milne (2003), Long patterns in the North Atlantic and the Mediterranean Sea, Geophys. Res. wavelength sea level and solid surface perturbations driven by polar ice Lett., 34, L18604, doi:10.1029/2007GL030641. mass variations: Fingerprinting Greenland and Antarctic ice sheet flux, Marcos, M., F. M. Calafat, W. Llovel, D. Gomis, and B. Meyssignac Space Sci. Rev., 108,81–93. (2011), Regional distribution of steric and mass contributions to sea level Vestøl, O. (2006), Determination of postglacial land uplift in Fennoscandia changes, Global Planet. Change, 76, 206–218. from leveling, tide-gauges and continuous GPS stations using least McClimans, T. A., B. O. Johannessen, and T. Jenserud (1999), Monitoring squares collocation, J. Geod., 80, 248–258. a shelf edge current using bottom pressures or coastal sea-level data, Wahl, T., J. Jensen, T. Frank, and I. Haigh (2011), Improved estimates of Cont. Shelf Res., 19, 1265–1283. mean sea level changes in the German Bight over the last 166 years, McDougall, T. J. (1987), Neutral surfaces, J. Phys. Oceanogr., 17, 1950–1964. Ocean Dyn., 61, 701–715. Meehl, G., et al. (2007), Global climate projections, in Climate Change 2007: Wakelin, S. L., P. L. Woodworth, R. A. Flather, and J. A. Williams (2003), The Physical Science Basis, edited by S. Solomon et al., pp. 847–940, Sea-level dependence on the NAO over the NW European continental Cambridge Univ. Press, Cambridge, U. K. shelf, Geophys. Res. Lett., 30(7), 1403, doi:10.1029/2003GL017041. Meyssignac, B., F. Calafat, S. Somot, V. Rupolo, P. Stocchi, W. Llovel, and Walsh, J. E., W. L. Chapman, and T. L. Shy (1996), Recent decrease of sea A. Cazenave (2011), Two-dimensional reconstruction of the Mediterra- level pressure in the central Arctic, J. Clim., 9, 480–486. nean sea level over 1970–2006 from tide gage data and regional ocean Weiss, J., J. Overpeck, and B. Strauss (2011), Implications of recent sea circulation model outputs, Global Planet. Change, 77,49–61. level rise science for low-elevation areas in coastal cities of the contermi- Milne, G. A., J. L. Davis, J. X. Mitrovica, H. Scherneck, J. M. Johansson, nous U.S.A., Clim. Change, 105, 635–645. M. Vermeer, and H. Koivula (2001), Space-geodetic constraints on glacial Woodworth, P. L., and R. Player (2003), The permanent service for mean isostatic adjustment in Fennoscandia, Science, 291(5512), 2381–2385. sea level: An update to the 21st century, J. Coastal Res., 19, 287–295. Mitrovica, J. X., M. E. Tamisiea, J. L. Davis, and G. A. Milne (2001), Woodworth, P., W. Gehrels, and R. Nerem (2011a), Nineteenth and twen- Recent mass balance of polar ice sheets inferred from patterns of global tieth century changes in sea level, Oceanography, 24,80–93. sea-level change, Nature, 409, 1026–1029. Woodworth, P., M. Menéndez, and W. Roland Gehrels (2011b), Evidence Østerhus, S., and T. Gammelsrød (1999), The abyss of the Nordic seas is for century-timescale acceleration in mean sea levels and for recent changes warming, J. Clim., 12, 3297–3304. in extreme sea levels, Surv. Geophys., 32,603–618, doi:10.1007/s10712- Pardaens, A., J. Gregory, and J. Lowe (2011), A model study of factors 011-9112-8. influencing projected changes in regional sea level over the twenty-first Yin, J. J., M. E. Schlesinger, and R. J. Stouffer (2009), Model projections of century, Clim. Dyn., 36, 2015–2033. rapid sea-level rise on the northeast coast of the United States, Nature Pavlov, V. (2001), Seasonal and long-term sea level variability in the Geoscience, 2, 262–266. marginal seas of the Arctic Ocean, Polar Res., 20, 153–160. Yin, J. J., S. M. Griffies, and R. J. Stouffer (2010), Spatial variability of sea level rise in twenty-first century projections, J. Clim., 23, 4585–4607.

12 of 12

! Intervals of likely sea level rise in a 100 years perspective, in cm relative to the shore.